Untitled

Transcript

Signal Data Explorer User Manual Version 1.2 21st April 2011 Author: Grant Brewer [email protected] www.signaldataexplorer.com Copyright Cybula Ltd. 2011 Disclaimer This document is protected by copyright and distributed under licences restricting its use, copying, distribution and decompilation. No part of this document may be reproduced in any form by any means without prior written authorisation of Cybula Ltd. Documentation is provided “as is” and all express or implied conditions, representations and warranties, including any implied warranty of merchantability, fitness for a particular purpose or non-infringement, are disclaimed, except to the extent that such disclaimers are held to be legally invalid. 1 Contents 1 Installation and Getting Started 5 1.1 Installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.2 Starting SDE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.3 Loading data files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.3.1 Loading a data set using the pull-down menu . . . . . . . . . . . . . . . . 8 1.3.2 Loading a data set using the right-click menu . . . . . . . . . . . . . . . . 8 1.3.3 Loading a data set using command line parameters . . . . . . . . . . . . . 8 1.3.4 Opening local data using drag-and-drop . . . . . . . . . . . . . . . . . . . 9 1.3.5 Loading a remote data set . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 1.3.6 Loading a zipped data set . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 1.3.7 Loading the MCD data format . . . . . . . . . . . . . . . . . . . . . . . . 10 1.3.8 Loading the Neuroshare data format . . . . . . . . . . . . . . . . . . . . . 11 1.3.9 Loading text data into SDE (CSV files) . . . . . . . . . . . . . . . . . . . 12 1.3.10 Inserting sample times into CSV data . . . . . . . . . . . . . . . . . . . . 12 Changing the settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1.4 2 Exploring the Data 16 2.1 Browsing the data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2.2 Displaying and playing multiple features in the main view . . . . . . . . . . . . . 18 2.2.1 Adding a feature to the view . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.2.2 Deleting a feature from the view . . . . . . . . . . . . . . . . . . . . . . . 18 2.2.3 Changing and re-arranging sub-views . . . . . . . . . . . . . . . . . . . . . 21 2.2.4 Scaling and display options . . . . . . . . . . . . . . . . . . . . . . . . . . 21 2.2.5 Saving the current settings . . . . . . . . . . . . . . . . . . . . . . . . . . 21 2.3 Saving selected variable data to a file . . . . . . . . . . . . . . . . . . . . . . . . . 21 2.4 Adjusting the Variable Selection List . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.5 Printing and exporting the view . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2 3 The Auxiliary View Window 3.1 24 Manipulating the auxiliary views . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.1.1 Resizing the auxiliary window . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.1.2 Zooming in and out . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.1.3 Coordinates, grids and time stamps . . . . . . . . . . . . . . . . . . . . . 26 3.1.4 Measuring the values of the displayed variables . . . . . . . . . . . . . . . 26 3.1.5 Changing the signal view scaling and vertical offset . . . . . . . . . . . . . 26 3.1.6 Superimposing multiple views . . . . . . . . . . . . . . . . . . . . . . . . . 27 3.1.7 Viewing more than one set of data using decoupled views . . . . . . . . . 28 3.1.8 Saving the auxiliary view as a bitmap . . . . . . . . . . . . . . . . . . . . 28 Applying filters to raw data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 3.2.1 The filter definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 3.2.2 Visualisation of the filtering output . . . . . . . . . . . . . . . . . . . . . . 32 3.2.3 Using the filtered data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 3.2.4 Applying filters to multiple views . . . . . . . . . . . . . . . . . . . . . . . 33 3.3 Detect spikes and save spike time data to file . . . . . . . . . . . . . . . . . . . . 33 3.4 The multiple channel array view . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 3.2 4 Searching for Similar Patterns 4.1 4.2 4.3 36 Searching the current dataset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 4.1.1 Selecting the ROI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 4.1.2 Invoking the search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 4.1.3 Search measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 4.1.4 The search and results view . . . . . . . . . . . . . . . . . . . . . . . . . . 40 4.1.5 Noise suppression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 4.1.6 Saving feature patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 4.1.7 Using the clipboard data . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 The Pattern Finder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 4.2.1 Importing a pattern to the pattern view window . . . . . . . . . . . . . . 44 4.2.2 Recalling a pattern from the pattern library . . . . . . . . . . . . . . . . . 44 4.2.3 Using the Draw Pad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 4.2.4 Adding a pattern to the pattern library . . . . . . . . . . . . . . . . . . . 44 4.2.5 Deleting a pattern from the pattern library . . . . . . . . . . . . . . . . . 45 4.2.6 Using the Pattern Finder to search the current data set . . . . . . . . . . 45 Searching for a single pattern from a feature database . . . . . . . . . . . . . . . 45 3 4.3.1 Invoking a database search . . . . . . . . . . . . . . . . . . . . . . . . . . 46 4.3.2 Manipulating the search tabs . . . . . . . . . . . . . . . . . . . . . . . . . 46 4.3.3 Viewing the search results . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 4.3.4 Variable names . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 5 The Task Planner 5.1 5.2 52 The task view . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 5.1.1 Creating a new task . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 5.1.2 Saving and loading tasks to/from a library file . . . . . . . . . . . . . . . 55 The results view . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 5.2.1 Using “Search” to perform a multiple pattern search on historical data . . 55 5.2.2 Using “Find” to perform a multiple pattern search on the current data set 56 5.2.3 Using the right-click menu of the results view . . . . . . . . . . . . . . . . 6 The SDE Protocol 6.1 6.2 57 59 Actions and data sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 6.1.1 The action field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 6.1.2 The data source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 6.1.3 The option field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 Examples of using the SDE protocol . . . . . . . . . . . . . . . . . . . . . . . . . 60 6.2.1 Starting SDE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 6.2.2 Closing SDE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 6.2.3 Opening a remote data file . . . . . . . . . . . . . . . . . . . . . . . . . . 61 6.2.4 Searching for a pattern 61 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 The Data Builder 62 7.1 The variable configuration file . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 7.2 Building a feature database using the command line interface . . . . . . . . . . . 63 7.3 Building a feature database from inside SDE . . . . . . . . . . . . . . . . . . . . 65 A The SDE Licence Manager 67 A.1 Purchasing SDE Standard or Professional Edition . . . . . . . . . . . . . . . . . . 67 A.2 Upgrading or Renewing your Licence . . . . . . . . . . . . . . . . . . . . . . . . . 69 A.3 Migrating your licence to a new machine . . . . . . . . . . . . . . . . . . . . . . . 69 A.4 Extended Trials, Multiple Licence Discounts and Permanent Licences . . . . . . . 69 A.5 Updating SDE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 B Shortcut Keys 70 4 Chapter 1 Installation and Getting Started Signal Data Explorer (SDE) is a general purpose data browser and search engine for time-series signal data. It assists in the visualisation of multiple signal data channels and enables the user to examine the details of one or more features extracted from the time-series data. SDE also contains efficient spike detection and dynamic filtering functionality. SDE’s efficient search engine allows a user to search for user defined patterns amongst either the loaded data or a pre-compiled data library. Users can maintain a library of extracted features and search for multiple, different, coincident patterns, across an array of different data channels. 1.1 Installation SDE is provided as a Windows application, with its own installer entitled SDE Setup.msi and is packaged with Microsoft Visual C++ 2008 Redistributable, which is required for SDE to function correctly. To begin the setup process, double-click on Setup.exe. SDE first checks to see if the Visual C++ Redistributable package is currently installed and, if required, will guide the user through the installation process. Once the Visual C++ Redistributable is installed, follow the on-screen instructions to install SDE. During the SDE installation process, the user must select an installation directory and may also choose to have a shortcut placed on the Windows desktop. The data set to be examined should either be copied to a known location on the machine or be accessible via an attached network. To uninstall or repair the application use Add/Remove Programs from the Windows control panel or double click on the SDE Setup.msi file used in the installation process. To upgrade SDE to a new version, double click on the new SDE Setup.msi file provided on the SDE media. This will prompt the user to uninstall the old version before installing the new version. In order to open Neuroshare data, SDE needs a set of dynamic link libraries that cannot be packaged directly for copyright reasons. The following files can be downloaded freely from the internet or can be copied directly from the SDE installation media. Copy the libraries into the Plugins→NsLib folder in the SDE installation directory. Once the files have been added, the “Files of Type” list in the open dialogue window will contain the Neuroshare data formats. • NeuroExplorerNeuroShareLibrary.dll 5 • • • • • • 1.2 nsAOLibrary.dll nsCedSon.dll nsNEVLibrary.dll nsNSNLibrary.dll nsPlxLibrary.dll nsTDTLib.dll Starting SDE SDE can be started by double-clicking on the application icon located in the start menu, the SDE folder or the Windows desktop. Alternatively, a command prompt window can be used. Figure 1.1 shows the opening screenshot of SDE; at the top of the window, there is a menu bar, a tool bar and a parameter bar. The main view window is located on the right and is used to display the time-series data in a variety of forms. The left of the screen is split into three areas: • The message window displays the messages relating to data processing and operations inside SDE. This can be cleared by selecting [Clear Message Window] from the [Edit] menu. • The results window lists the pattern search results and enables the user to select which results to view. By double clicking on a result in the results panel, the actual shape of the identified pattern can be viewed. • The control panel allows users to select features to be displayed, invoke searches and control the data being viewed. 1.3 Loading data files There are four ways to select the input data set to be loaded: (i) from the pull-down menu in the menu bar, (ii) from a right-click menu, (iii) drag-and-drop the data file onto the main view window or (iv) by using command line parameters. SDE can also download and open data from a remote site. The data formats currently supported by SDE are: • • • • • • • • • Multi-channel systems data (.mcd). Comma delimited (.csv). Alpha map (.map). Neural event (.nev). NeuroShare native (.nsn). Nex (.nex). PC spike2 (.smr). Plexon data (.plx). TDT data format (.stb). 6 Figure 1.1: The initial SDE screen with no data set loaded 7 If the format of the selected input data can be understood by SDE and there is no channel selection required, the features present in the file will be available for selection in the control panel and a section of the data will be displayed in the main view window. However, if the data opened is in Multi-channel systems or Neuroshare data formats, the user must first select the channels to be opened, using the channel selection dialogue window explained in sections 1.3.7 and 1.3.8. If the selected file cannot be understood, the message “Invalid file format” will be displayed in the message window. The following table shows the size limitations for opening data files in this version of SDE. However, note that these limitations only apply to the opening and exploration of data - SDE can create and search databases of many terabytes in size. Data format Comma delimited Multi-channel systems data Neuroshare data formats 1.3.1 Size limitations 2GB file size 2GB per channel 2GB per channel Loading a data set using the pull-down menu Select [Open] from the [File] menu to invoke the file open dialogue window. From this dialogue window, select the filename and file type and click the Open button. Alternatively, a recently used data file can be selected from the [Recent Files] list in the [File] menu. 1.3.2 Loading a data set using the right-click menu In SDE, a collection of frequently used menu items can be accessed from the right-click menu. This menu is context sensitive - organised by the current status of SDE. To load a data set from the right-click menu, click the right mouse button when the mouse indicator is in the upper part of the main view window. A floating menu is then displayed; select [Open] to invoke the file open dialogue window. From the dialogue window, select the file type and filename and click the Open button to open the input file in the SDE main view. Alternatively, click on [Recent Files] in the floating menu to select and open a recently used data set. 1.3.3 Loading a data set using command line parameters A set of command line parameters is provided to start SDE with a data set loaded at a specified position within the file. These parameters are: -p filename Load the data set described by the named file in the SDE main view window. -t hh:mm:ss Load the data set at time position hh:mm:ss. -r xxxxxx Load the data set at the xxxxxx position (in seconds) from the beginning of the data file. 8 The -t or -r option must be used together with the -p option. If the user includes both the -t and -r options, only the last occurrence of -t or -r will be used. If the -t or -r option indicates a position beyond the end of the current data, the view will show the data at the end of the current set. Similarly, if the -t option indicates a position that precedes the beginning of the data then the view will show the data at the beginning of the current set. Example: SDExplor -p myfile.csv -t 12:28:03 This will load the data set contained in the file myfile.csv and position the data view to begin at the time 12:28:03. SDExplor -p myfile.csv -r 1000 This will load the data set contained in the file myfile.csv and position the data view to begin at the 1000th second from the beginning of the file. 1.3.4 Opening local data using drag-and-drop SDE supports drag-and-drop to open local data sets. Any supported data format, including their zipped format, can be dragged to the main view window and opened by SDE. A forbidden sign will be displayed when an unsupported file is dragged onto SDE. Otherwise, a plus sign ‘+’ is displayed. If the data set is zipped, SDE will always first unzip it before checking whether it is a supported format. 1.3.5 Loading a remote data set SDE can open files from remote sites using the HTTP, FTP or SRB protocols. To open a remote data set, select [File] from the menu bar and then click [Open Remote File]. From the pop up window, type the remote URL and click on the Open button. SDE will automatically identify the data type and, once the download is complete, the file will be opened if it is a supported data format. By default, SDE will open the remote data immediately after the download has completed. However, this can be overridden by unchecking the [Open file when finish downloading] check box in the download pop up window, as shown in figure 1.2. If the remote data is zipped and consists of more than one file, a message window will pop up asking the user to select the data set to be opened. 1.3.6 Loading a zipped data set SDE can open zipped versions of all supported data types. This can help to reduce the quantity of data transferred through the network or reduce space on the local disk. The extension name of the zipped data file is not relevant as SDE will identify the data type inside the zip archive and use an appropriate engine to open it. An archive can contain multiple sets of data. If the data is in the TDD data format, the zipped file can also use the same .tdd extension name. 9 Figure 1.2: The internet download window In this case, a user can double click the zipped TDD file to open it, exactly as if opening an unzipped TDD file. 1.3.7 Loading the MCD data format SDE supports the importing of continuous raw data and spike time data from an MCD data file. MCD data can be selected to be loaded in the same manner as any other accepted data format (see section 1.3). However, when MCD data is opened in SDE, a dialogue box is displayed that allows the user to view the meta-data of the MCD file and to select which channels should be extracted and displayed. Figure 1.3 shows the MCD data file opening dialogue box, with all channels selected for extraction. Figure 1.3: A dialogue box for opening MCD data files 10 1.3.8 Loading the Neuroshare data format SDE can extract and display the information contained in two of the entity types that make up a Neuroshare data file: the analog and the neural event entities. Analog entities, like most of the data displayed in SDE, are shown as continuous lines, whilst the position of each neural event is identified by the presence of a vertical line at the appropriate position on the time line. Event and segment entities cannot be viewed in this version of SDE. When a Neuroshare data file is opened, a dialogue box will appear displaying information about the contents of the file and enabling the user to select the elements of the data to be viewed. The channel selection dialogue window, shown in figure 1.4, contains two drop-down lists “Stream Type” and “Entities”) and a “Channels to View” selection panel, which are used to ensure that only data produced with an identical parameters set can be opened simultaneously inside SDE. Figure 1.4: A dialogue box for opening Neuroshare data files To select the data for viewing, first pick the appropriate entity type from the “Stream Type” drop-down list. All the entities of that type, contained in the data file, will then be displayed in the “Entities” drop down list. Selecting a particular channel from the “Entities’’ list will cause the “Channels to View” panel on the right of the window to display all the channels that were generated using an identical parameter set to that channel. For example, consider a data file that contains six analog entities: a1, a2 and a3, which have one parameters set and b1, b2 and b3, which were produced with a different parameter set. If a1 is selected in the “Entities” list, only a1, a2 and a3 will be available for selection in the “Channels to View” selection panel. Once the entity has been selected, the user can select the channels to be opened from the “Channels to View” selection panel. By placing these restrictions on the selection panel 11 contents, the user is prevented from attempting to open a set of channels that have been produced using different parameter sets and which can therefore not be displayed synchronously in SDE. If a user wishes to simultaneously examine data produced using different parameter sets, decoupled Auxiliary View Windows (explained in section 3.1.7) can be used. Channels can be selected or deselected in the “Channels to view” list by clicking on the appropriate check box. Alternatively, the shortcuts listed below can be used to add or remove multiple channels. To select multiple list elements simultaneously, select the required channels with the Ctrl Key depressed or utilise the Shift key to select a range of channels. Ctrl + A Ctrl + U Ctrl + S Ctrl + R Check all items Uncheck all items Check all selected items Uncheck all selected items Once the required channels have been selected, click Open, or press the enter key to open and view the requested data. Alternatively, click the Cancel button to close the dialogue window without opening a new data set. 1.3.9 Loading text data into SDE (CSV files) Although text data requires a greater amount of disk space and takes longer to load than the binary formats, it is often the simplest way to get data into SDE. Text data can also be viewed and edited much more easily than binary data, using any standard text editor. At its simplest, the first line of a CSV data file contains the name of each variable, separated by commas. The second line contains the values for each of the variables at the first time interval, also separated by commas, with each subsequent line containing the comma separated values for the subsequent time interval. An example of the start of a CSV file is shown below, containing three variables: the first is increasing in value, the second decreasing and third remaining unchanged. For further examples of CSV files, examine the sample data available on the SDE media or website, or create your own by saving sections of other files from within SDE. Increasing Variable,Decreasing Variable,Unchanging Variable 200,86.2,-12 210,80.1,-12 220,67.0,-12 230,50.24,-12 Details on how to insert timing information into CSV files will be discussed in section 1.3.10. 1.3.10 Inserting sample times into CSV data Whereas MCD and Neuroshare data formats have well defined methods for storing information about sampling rates and sampling times, this is not the case with CSV data. Therefore SDE 12 provides two methods for attaching sample times to the data, so that they can be read into SDE: using a timestamp and in a file header. If no timing data is incorporated in the file, SDE will assume a sample rate of 1Hz and a start time of 00:00:00.0. To use timestamps, create an additional column at the left hand side of the data and set the variable name to “TIMESTAMPS”. Set the value of that column to the time (in seconds) of each sample point. The timestamps will not be plotted as a variable, but will define the time that corresponds to each data point loaded into SDE. Note that this version of SDE plots each data point linearly and therefore, if the timestamps are not equidistant, the shape of the plotted data may be misleading. However, the times of each point will still be reported correctly. Timing information can also be inserted into CSV files using a header, placed between the tags and . The first line in the header defines the file type and must be either “Type: Even Time Series” or “Type: Spike Time”. If the first line of the header does not specify one of these file types, the entire header will be ignored. If both timestamps and a file header are used, the timestamps take precedent. An example of an even time series header can be seen below. It shows the notation that can be used to specify the length, start time, end time and sampling rate of the data. The minimum information required to fully specify the time, at every point in the data, is the start time and sampling rate. Type: Even Time Series Data Length: 2500000 Start Time: 2008-10-8*02:46:40.00 End Time: 03:04:05 Sampling Rate: 250 (Hz) An example of a spike time header can be seen below. Source Data Length and Source FileName are the length and name of the analogue time series raw data. Each variable in the spike time file consists of a list of the times of each spike in data points and the Time Resolution is the length of time, in seconds, to which each data point corresponds. Type: Spike Time Source Data Length: 1488000 Start Time: 2006-12-04*12:57:05 Source FileName: Blue Light 2.mcd Time Resolution: 0.000100 Further examples of both Spike Time and Even Time Series files can be obtained by saving such data types inside SDE. 13 1.4 Changing the settings Both the current and default SDE settings can be changed in the Settings dialogue window, which can be invoked by selecting [Settings] from the [Preferences] menu. As some of the settings are data dependent, this menu item is enabled only after a set of data has been opened in SDE. The Settings dialogue window can be seen in figure 1.5 and is used to set the current display parameters and to control the behaviour of the SDE search engine. It should be noted that the search parameters specified in the settings page apply only to searches performed using the main search button, located on the control panel. The pattern finder and task planner (discussed in section 4.2 and chapter 5 respectively) provide alternative methods for the specification of search parameters. Figure 1.5: The display and searching parameters setting page • Upper bound and Lower bound define the range of the scale used to display the current data set in the bottom right feature window. • Threshold is a parameter that defines the threshold when pattern searching. Only matching scores higher than the threshold are displayed on the results window. An exact match is defined as 500, so setting a threshold of 480 would return only closely matching patterns, whereas a threshold of 50 would deliver a much broader set of matched results. • The Match Low and Match High parameters define an amplitude window, which constrains a similarity based search, by removing any matching patterns whose amplitude does not fall within the specified window. The amplitude of a pattern is defined as the mean 14 of the sum of its absolute values and the amplitude window is calculated by multiplying the amplitude of the template pattern by the Match Low and Match High values. • The Buffering data from network disk check box defines whether or not to buffer data from a mounted network disk. Buffering from a network disk can improve the performance when playing the data but increases the use of local disk space. There is an automatic check of available local disk space and a pop-up warning appears if there is insufficient disk space available. • The Library Directory defines the default searching library directory. SDE enables users to build a library of data records, which can be searched to locate any data sections that match a selected pattern. When a search is invoked from SDE, the default library from the default library directory is searched unless an alternative is explicitly defined. • Default URL defines the default URLs for invoking pattern searching over a network. The number of the default URL is defined by the available, dynamic plug-ins. Each of the plug-ins can have one default URL. The URL in the SDE address window changes to the default setting once a user selects the search engine from the pull-down menu. • Set as default saves all the input parameters from the Settings Window as defaults if the check box is ticked when the OK button is pressed. 15 Chapter 2 Exploring the Data Figure 2.1 shows SDE with a data set loaded. The main SDE window consists of two parts: the control panel on the left side of the window and main view on the right. The upper part of the main view window is the data view, which displays data from multiple channels simultaneously. The feature view at the bottom of the main view window shows data from one particular channel and is used to select the data segments that will be used as search templates. 2.1 Browsing the data Once the data file is opened, it can be displayed and investigated using a variety of different working modes. Data can be “played” against time using the play control buttons located on the lower right side of the control panel, as shown in figure 2.2. The four blue play control buttons on the control panel are, from left to right, play/stop, backwards, forwards and rewind, respectively. The play/stop button is context sensitive and appears as a play button when playing is stopped or as a stop button when data is being played. When playing data, clicking the play/stop button stops the playing and, when stopped, clicking the play/stop button starts playing. The arrow on the play button indicates the direction of the current play; clicking the backward or forward button will change the direction. When data playing is stopped, clicking the forward (or backward) button moves the data forward (or backward) by one frame. The speed control slide bar allows the user to control the play speed, by dragging the slide bar to the left or right. To control the speed with greater precision, highlight the slide bar by clicking on it and then use the left or right arrow keys to increase or decrease the play speed. As the size of the datasets than can be opened in SDE can vary hugely, SDE has two speed bands to help the user set the speed to an appropriate value. In the lower speed band the speed can vary between 1 and 250, whilst in the higher speed band it ranges from 0 to 10 000. To switch from the lower to the higher speed band, move the slider to the right hand end of the slide bar and confirm that you would like to change band. To switch from the higher to the lower speed band, move the slider to the left hand end of the slide bar and confirm the change. At the bottom of the control panel, the play control slide bar indicates the position of the current view in the loaded data set. The view can be moved to any desired point by moving the indicator on the slide bar or by clicking on the desired position. For example, to go to the end 16 Figure 2.1: The SDE screen with a data set loaded of the data, click the right end of the play control slide bar. To go to the beginning of the data, click the rewind button or the left end of the play control slide bar. The exact time or data point number of the current view can be found on the parameter bar, near the top of the window, in the T offset parameter box. The T Offset parameter box can be made to display either the time or the number of data points by selecting [View/T offset] from the main menu and selecting either [Show Time] or [Show Points]. Alternatively, the T key can be used to switch between displaying point and time offset. To move to an exact point or time offset, type the appropriate value into the T Offset parameter box and press the Return key. A means of precisely locating the time and value of each variable, at any point on the display, is provided through a vertical line called the time instant indicator, which appears if the middle mouse button is clicked anywhere within the data view window. Alternatively, this can be done by holding down ALT and using the left mouse button. The T offset box shows the time at that 17 Figure 2.2: The SDE control panel instant (in points or time, as selected) and the value of each variable at that time is displayed at the top of each variable view. Pressing the middle button, when over the indicator, and moving the cursor shifts the time instant indicator continuously and the T offset box value is updated correspondingly. To turn off the time instant indicator, move it to the left of the main view window until it disappears. 2.2 Displaying and playing multiple features in the main view Figure 2.3 shows SDE with multiple, synchronised sub-views displayed in the main view window. Each sub-view lists the name of each channel and the value of that variable at the time indicated by the time instant indicator bar, discussed in section 2.1. This provides a useful method of precisely measuring the data values. 2.2.1 Adding a feature to the view New variables can be added to the main data view by using the right-click menu and selecting [Multiple View/Add View]. An add view dialogue window appears and all available variables are listed in the drop down list box, as shown in figure 2.4. Choose a variable and press the Add button to add a sub-view to the main view window. Multiple variables can be selected by using CTRL + LEFT CLICK to select subsequent channels. A block of variables can be selected by left clicking to define the beginning of the block and using SHIFT + LEFT CLICK to define the end. There are two further buttons to the right of the Add button. The first is Sort/Unsort, which toggles the order in which the variables are listed, between alphabetical and the original channel order. To the right of this, the Load All button removes any restrictions on the variables that are displayed. For details on how to restrict the number of variables displayed in the list, to make the location of the required channels easier, consult section 2.4. 18 Figure 2.3: Multiple feature view 19 Figure 2.4: The add view dialogue window 20 2.2.2 Deleting a feature from the view A sub-view can be removed from the multi-feature view by first left clicking inside the sub-view to activate it (the colour of the feature image becomes red). Then use the right-click menu and select menu item [Delete View] or select [Delete All Views] to remove all views from the data view window. Alternatively, a sub-view can be deleted by left clicking the sub-view whilst holding down the DELETE key. 2.2.3 Changing and re-arranging sub-views Pressing the RIGHT/LEFT keys changes a sub-view to the next/previous feature in the channel list. Alternatively, by selecting the Enable View Scroll option from the Multiple View submenu of the right click menu, a scroll bar is displayed on the right side of the main view window. This can be used to change the variables displayed in the data view window. To rearrange the order of sub-views in an auxiliary window, drag and drop the sub-view to the preferred position. 2.2.4 Scaling and display options When a sub-view is added to the data view window, SDE automatically calculates the scaling factor and the upper and lower boundaries to ensure that all data values of that signal are within the view. However, the scaling factor of the active sub-view can be decreased or increased using the Up or Down keys. The upper and lower boundaries of the view (the y-offset) can be shifted up or down by using the Shift + Up or Shift + Down key combinations. The horizontal time scale can be hidden or displayed on the main view window by checking or unchecking the [Hide Time Stamps] option in the [Tools] menu. 2.2.5 Saving the current settings The multiple feature view settings define the default features that will be displayed in the main data views when a data set is opened. To save the multiple channel view settings, open the pull-down menu [View] and select the menu item [Save Settings]. The settings will be saved to the application configuration file - only one set of multiple feature view settings may be saved. When the first set of data is loaded, multiple variables are automatically added to the main view window, according to the default settings. When subsequent data sets are opened, the multiple channels view is loaded using the previous settings. However, the default settings can be recalled by pressing the U key. 2.3 Saving selected variable data to a file SDE enables a user to select a section of data and a subset of the available variables and save this data as a new CSV file. This can be done by selecting [Save Data] from the [File] pull-down menu to invoke the save data dialogue window shown in figure 2.5. Select the channels that are 21 to be saved, type in a filename and then click OK. The multiple channel data is saved in the specified file together with the appropriate time stamp values. Figure 2.5: Saving a selected data segment Channels can be selected or deselected in the “Check variables to save” list by clicking on the appropriate check box. Alternatively, the shortcuts listed below can be used to more easily add or remove multiple variables. To select multiple list elements simultaneously, select the required channels with the Ctrl key depressed or utilise the Shift key to select a range of variables. Ctrl + A Ctrl + U Ctrl + S Ctrl + R Check all items Uncheck all items Check all selected items Uncheck all selected items The user can choose to either save the displayed multiple channel data for a specific time interval or to save it for the complete time interval contained in the currently loaded data set. These options are chosen as follows: • If a time interval is selected prior to performing the save operation, then data for each of the multiple variables selected is saved, but only for the specified time interval. The time interval must be selected (highlighted in blue) in the feature window (the lower part of the SDE main window view) or in a Auxiliary View Window. To select a section of data in the feature window, left click at the beginning of the desired region and drag the mouse button to the right. Navigate through the data until the end point is visible in the main view window and then click on the end point with the right mouse button. Alternatively, click and drag over the region using the middle mouse button. 22 To highlight a section of data in the Auxiliary View Window, hold down the Shift button and select a region of the data using the middle mouse button. Press Escape to cancel the selection. For more details on selecting a region of interest see section 4.1.1. • If a time interval is NOT highlighted in the lower part of the SDE main window, then all the selected variable data (corresponding to the time interval for the whole data set) is saved. 2.4 Adjusting the Variable Selection List When a dataset is opened in SDE all the variables present in the data are listed in the Control Panel in the order in which they were arranged in the file. Clicking on any variable in the list causes the data to be displayed in the feature view window, where sections of the data can be selected to be saved as a new datafile or used as a search template. However, when the dataset contains large number of variables, the required variable can often be difficult to locate. By right clicking on the Variable Selection List, the user is provided with a set of options that organise or filter the variable list, so that the required data can be located more easily. Furthermore, the adjustments made to the Variable Selection List in the Control Panel will also affect the variables listed when adding views to the main view window or AVW. • Sort by Names. Sorts the Variable Selection List into alphabetical order. This option is only available when the list is currently unsorted. • Unsort Variable List. Sorts the Variable Selection List according to the position of each variable in the original data file. This option is only available when the list is currently sorted alphabetically. • Select Variable Subset. Produces a new dialogue window containing a check list of all the variables currently loaded into SDE, which can be used to select the variables that are included in the Variable Selection List. Selected the required variables and press the Set button to apply the restrictions to the variable list or press Cancel to leave the list unchanged. To select a range of variables, click on the required channels with the Shift key depressed. Then press Ctrl + S to check all the selected variables or Ctrl + U to uncheck them. Other useful shortcuts are Ctrl + A, which checks all the variables in the list and Ctrl + U which unchecks them all. • Load All Variables. Removes any restrictions on the variables selected and displays all of the variables loaded into SDE in the Variable Selection List. The option is only available when there is currently a restriction in place on the variables displayed. 2.5 Printing and exporting the view To print the contents of the main view window (both top and lower parts), select [File] from the menu bar and choose [Print]. The contents of the data view window can be saved to the 23 clipboard, as a bitmap image, by selecting [Copy] from the right click menu of the main view window. 24 Chapter 3 The Auxiliary View Window In addition to the main view window, SDE provides an Auxiliary View Window (AVW) for dynamic visualisation of multiple time-series data. By default, the auxiliary views are synchronised and coupled with the main view, but they can also be decoupled to allow independent exploration of different time segments. Once an AVW has been decoupled from the main window, it cannot be re-coupled. The auxiliary view provides a range of additional options for exploring the time-series data. From the AVW, a user can: • • • • Resize the window Condense or magnify the data Change the scaling factor and offset of the signal in each channel View channels with different zoom factors and display more than one dataset on different decoupled auxiliary views • View superimposed variables • Apply filters to one or more variables • Perform spike detection AVWs are initiated by selecting [Aux-Window/New Aux-Window] from the main view right-click menu, or by using the Alt+N short-cut, key combination. On an AVW, a sub-view can be activated by clicking on its sub-window using the left mouse button. The plotted data changes to red to indicate that it is activated. Unless otherwise specified, all operations that are initiated from right-click menu of an AVW, or from the keyboard, are applied to the active sub-view only. An example of an AVW is shown in figure 3.1. 3.1 Manipulating the auxiliary views By default, a new AVW inherits the sub-views from the data view on the main window. However, sub-views can be added, deleted and re-arranged using the right click menu, in the same manner used to manipulate the main view window (see section 2.2). If the number of sub-views present in the AVW is less than the number of data channels loaded into SDE, the view scroll bar will 25 Figure 3.1: An auxiliary view window automatically be enabled, allowing the user to adjust the channels displayed. If the user has performed any processing on a sub-view, such as the application of a filter or spike detection, the altered view will not be scrolled out of the window when the view scroll bar is used. This prevents the loss of the processed data. Just like the main SDE window, the auxiliary view windows have a vertical, yellow, time instant indicator and all the indicators are synchronised at the same data point. Moving one time instant indicator causes all the other indicators and data views to adjust automatically to maintain this synchronisation. If the time instant indicator in a window is not set, the left edge of the view is synchronised with the other indicators. The current, synchronised, time instance (the offset from the start of the data in points) is displayed in the T-Offset text box on the control panel of the auxiliary view. The user can shift the time instant indicator, by clicking on the AVW with the left mouse button, whilst holding down the ALT key, or by using the middle mouse button. Alternatively, the user can adjust the view, without changing the synchronised data point, by holding down the CTRL key whilst moving the indicator with the middle mouse button pressed. 3.1.1 Resizing the auxiliary window The size of an AVW can be adjusted up to the size of the desktop working area. There are two modes of resizing: the normal mode and the cropped view mode. The default setting is the normal mode, which resizes all sub-views proportionally as the AVW is changed. In contrast, in the cropped mode, the size of the sub-view is not changed when the AVW is resized. To set the AVW to cropped mode, check the right-click menu item [Resize as Cropped View]; uncheck it to return back to the normal mode. 26 The resizing mode changes how the layout of the window is affected by adding or deleting a sub-view from an AVW. In normal mode, the size of the window remains unchanged whilst all the sub-views adjust their sizes to fit the auxiliary window. In cropped view mode, the size of all sub-views remains unchanged, whilst the auxiliary window changes its size to accommodate all the sub-views until the height of the auxiliary window is equal to the height of the desktop working area. 3.1.2 Zooming in and out Zoom in/out can only be applied to the AVW and allows a user to view a much longer time span within a single window, or zoom in to more closely examine a small section of the data. The zoom factor can be set within the range of -25 to 2000. A positive value will cause the view to zoom out and compresses the data, whilst a negative value zooms in. When the zoom value is less than 1, there will be gaps between the data points. SDE fills in the gaps by linearly interpolating between neighbouring points, but plotted values will only be displayed when the time instant indicator is placed over one of the actual data points. To zoom in/out, simply click the spin-buttons on the control panel of the auxiliary view or type in the required zoom factor in the edit box next to the spin-buttons and press ENTER. 3.1.3 Coordinates, grids and time stamps The Y coordinate of each sub-view is dynamically set by SDE and a light grey grid can be plotted onto each of the sub-views. By default, SDE plots coordinates and a grid on each sub-view, but a user can turn these settings on or off. The settings are applied to all sub-views on an AVW. From the right-click menu, select [Coordinates] and use the appropriate sub-menu items to reset the parameters. The grid can be turned on or off when the scale is plotted but, if the scale is turned off, the grid cannot be viewed. The time stamp can be displayed or hidden, in an AVW, by checking or unchecking the right-click menu item [Hide Time Stamps]. The font size of the time stamp can be changed (together with all the font sizes in the view) using the right-click menu [Coordinates/Font Size]. 3.1.4 Measuring the values of the displayed variables The top of each sub-view lists the variable name and its value at the synchronised time instance. Only the first 22 characters of the variable name are displayed in the sub-view. To view the full name of a variable, move the cursor over the variable name and press the left mouse button. 3.1.5 Changing the signal view scaling and vertical offset When a signal is opened in the AVW, SDE automatically calculates the scaling factor and the upper and lower boundaries to ensure that all data values of that signal are within the view. However, the scaling factor of the active sub-view can be increased or decreased using the Up or Down keys. The upper and lower boundaries of the view (the y-offset) can be shifted up or 27 down by using the “Shift+Up” or “Shift+Down” key combinations or by selecting [Quick Settings] from the right click menu and entering the required values. There are three operations, relating to the scaling factor settings, which are accessible from the right-click menu: 1. Unify Scaling Factor. This operation will apply an identical scaling factor and vertical offset to each sub-view. The unified scale will range from the minumum value present in any of the existing scales to the maximum value. 2. Apply As Global Scaling Factor. The scaling factor and vertical offset of the active sub-view is applied to all other sub-views. 3. Apply Scaling Factor to Sel. Views. The scaling factor and vertical offest of the active sub-view is applied to all other selected sub-views. As the default SDE scale settings ensure all data points are visible, a small number of outlying data points can greatly extend the scale, hiding much of the detail contained in the signal. However, by pressing the I key, SDE will adjust the scaling factor of the currently selected view, according to the data currently visible within the AVW. Therefore, by zooming in and adjusting the view such that the outliers are not part of the currently displayed data and using the I key, the user can scale the view according to the significant data points. 3.1.6 Superimposing multiple views Multiple variables can be displayed in the same view, by using the superimpose view option. First, click on the view to which you wish to add further variables. Then select the views that are to be added, by clicking on them with the Ctrl button pressed. Alternatively, hold Shift and click to select a range of consecutive views. The names of the additional channels selected will be highlighted in blue. Select [Superimpose Selected Views] from the Superimpose View menu, in the right click menu, and the additional channels will be plotted on top of the initially selected variable, as shown in figure 3.2. The current value and scale of only one of the superimposed variables can be shown in the view window and this is the variable highlighted in red. To change the highlighted variable, use the Left or Right arrow keys. If it is not clear which data points belong to each line, move the cursor over one of the points. A label will be produced containing the corresponding variable name. If you wish to accurately compare two variables using this method, you will need to ensure that they have identical scaling factors. Select [Apply Scaling Factor to Group] from the Superimpose View menu, (or use shortcut key G), to apply the scaling factor of the highlighted variable to all other variables present in the view. Alternatively, if you select [Recalculate Scaling Factor] (shortcut combination Ctrl+G), SDE will create a single upper and lower bound for all the variables in the view, which allows all the points from all the superimposed variables to be displayed. To remove the superimposition, select [Delete Superimposed View] from the Superimpose View menu, or select [Delete All Superimposed Views] to remove all superimposed views from the AVW. 28 Figure 3.2: Superimposed gait data for both feet, with the left foot variable highlighted 3.1.7 Viewing more than one set of data using decoupled views SDE can display more than one set of data simultaneously using multiple decoupled auxiliary view windows. To view two or more data sets, first open one set of data from the main view window. Then, invoke an auxiliary view window and select the sub-views of interest. From the right-click menu, uncheck the menu item [Coupled with main view]. SDE will ask for a confirmation of the decoupling operation. Select OK and the AVW will be decoupled from the main view and will not be automatically closed when a new data set is opened. Open the second set of data from the main view and invoke a second AVW for the new data set. The two sets of data can then be played, zoomed etc. separately on each AVW. A decoupled AVW cannot be re-coupled to the main view window. 3.1.8 Saving the auxiliary view as a bitmap An AVW view can be saved as a bitmap image. To save a view, use the auxiliary view right-click menu item [Save View], type in the file name and then click Save. When saving an AVW, three options are available for the customisation of the view. • Using white background changes the colour scheme to incorporate a more printer friendly white background. • If Using white background is selected, the user can also chose Black line drawing to create a completely colour free picture. • Thick line drawing increases the size of the plotted data lines. Figure 3.3 shows a saved AVW with Using white background and Black line drawing checked and with timestamps switched off. 29 Figure 3.3: A saved Auxiliary View Window 3.2 Applying filters to raw data SDE provides a set of filters for signal processing that can be applied to any variables displayed in the AVW. To apply one or more filters to a signal, select the desired feature and click on [Apply Filters] in the Filters submenu of the AVW right-click menu. A dialogue box, “Filter Settings Panel”, is invoked, as shown in figure 3.4. For performance reasons, filtering is currently unavailable on large datasets. If there is less than 1 GB of memory available and a single channel is greater than 500 000 bytes in size, the Apply Filters option will not be available for selection from the right-click menu. If there is more than 1 GB of memory available, this limit is increased to 1 000 000 bytes per channel. Figure 3.4: The Filter Settings Panel 30 The Filter Settings Panel is split into four sections: 1. Applied Filters, located in the top-centre of the panel. This view displays all the filters applied to the current view. The control button Delete can be used to delete the selected filter from this view. Filters that appear in this view are immediately applied to the active view. Users can view results and adjust the filter settings using the controls present in the parameters panel. 2. Available Filters, located underneath the applied filters panel, lists all the available filters. The selected filter from this combo box can be added to the applied filter list using the Add or Insert buttons. The Add button adds a new filter to the end of the applied filter list, whereas the Insert button inserts a new filter in front of the currently selected filter on the applied filter list. The filters currently available are Gaussian, Mean, Envelop, Median, Binary and Bi-limiter and Event Rate and are discussed in section 3.2.1. 3. Parameters edit boxes are located on the top-right side of the panel. When a filter in the Applied Filters panel is selected, the relevant parameters of that filter are shown in these boxes. The parameter names are context sensitive. The parameters can be changed either by using the spin-buttons or simply by typing the number into the edit box. The active sub-view will be changed immediately after the parameters have been changed. Keeping a spin-button depressed will repeatedly change the parameter value. The speed of the change can be adjusted by pressing the CTRL or ALT key whilst the clicking the spin-button. CTRL+spin button increases the speed by 10 times, whereas ALT+spin button decreases the speed by 10 times. The speed change operation only applies to floating point data input boxes. 4. Filter Set Library. The applied filters and associated settings can be grouped as a filter set and saved to a filter library for later use. When the applied filters and their parameters meet the user’s requirements they can be saved by clicking the control button Save. Type a suitable name into the input dialogue (see figure 3.5) and then click the OK button to save the new filter set. Use the Remove button to remove a filter set from the library. To apply the filter set to the current data click the Load button. After a filter set has been loaded into the Applied Filters panel it can be modified in the normal way. 3.2.1 The filter definitions Currently, eight types of filter are available: 1. The Gaussian Filter is a kernel based filter based on a Gaussian distribution function. Convolution is applied to the signal by using the Gaussian kernel. This filter is useful for smoothing the data. Only one parameter, sigma, is required. 2. The Mean Filter is also a kernel based filter, providing mean value plotting of the data. It is similar to a Gaussian filter, using less computation, but it is not as effective at smoothing a signal. The only parameter for a mean filter is the kernel Width. 31 Figure 3.5: Saving a filter set 3. The Envelope Detector, simulating an electronic signal envelope detector, provides a way of extracting the signal envelope from the raw signal. Parameter Tao defines the time delay used to keep a previously detected peak. The lower the Tao value, the more sensitive the filter is to changes in the envelope shape. The Boolean parameter Method allows the user to define which envelope bound is to be detected. Method=0 defines a lower-bound envelope detector and Method=1 defines an upper-bound envelope detector. 4. The Median Filter can be used to filter out typical ‘salt-and-pepper’ noise in signal processing. In time series data, this ‘salt-and pepper’ noise appears as spurs. A median filter sorts the data within a defined width of window and then uses the value at the centre of the sorted list to provide the output. Using this output, the spurs on the signal can be removed. The only parameter for a median filter is the window Width. 5. The Binary Filter thresholds a signal to generate a binary output. Four parameters need to be defined for a binary filter: • Threshold defines the threshold applied to the signal. • Boolean parameter method defines the method used to convert the data to a binary signal. When method=0, the filter uses negative logic (i.e. higher than threshold is logical 0, lower is logical l), whereas method=1 uses positive logic. • Parameter As mask defines the way the output is presented. When As mask=0, any data point exceeding the threshold is set to the threshold value, forming a square wave. If As mask=1, the raw data value is taken for any values that exceed the threshold. • The Alias value is the value assigned to any data points that do not exceed the threshold. 6. The Bi-limiter provides a means of getting rid of outliers and re-filling the signal values. There are two thresholds applied to the bi-limiter, the Up-threshold and the Lowthreshold. When the Up-threshold is higher than the Low-threshold, the valid signal value is in between the two values. Signal values higher than Up-threshold or lower than Low-threshold are treated as outliers. If the Up-threshold is lower than the Low-threshold, data values within the thresholds are treated as outliers. 32 Parameter method defines the way re-fill values are calculated to replace the outliers: • Method=0: a linear interpolation is used to replace values for the outliers. • Method=1: a value equal to the centre of the upper and lower thresholds is used to replace the outlier. • Method=2, values higher than the Up-threshold are set at the upper-threshold and values lower than Low-threshold are set at the value of the lower threshold. • Method=3 the Alias value parameter is used as the value of the outliers. 7. The Firing Rate filter gives a visual indication of the firing rate of spike time data, by displaying the reciprocal of the Inter-Spike Interval. The filter parameters refer to an alternative method of firing rate calculation that is currently in development. The Firing Rate filter currently operates on only raw spike time data and not on the SDE spike detector output. It cannot be meaningfully applied to continuous data. 8. The Difference filter is used to show the difference in value between two variables at each time instance. The Secondary Var. is used to select the variable to which the currently active variable should be compared and the Syn. Offset allows each point along one variable to be compared to a point on the secondary variable at a different relative position. For example, a Syn. Offset of 0 causes the two variable to be compared at identical time instances, whilst a value of 10 causes each point in the filtered variable to be compared to the point 10 time instances earlier in the secondary variable. The secondary variable is multiplied by the Coefficient parameter before it is subtracted from the primary variable. For example, a Coefficient of 1 is a straight subtraction, whilst a Coefficient of -1 causes the two variables to be added together, instead of subtracted. 3.2.2 Visualisation of the filtering output When selecting the required filters and parameters, the output of the specified filtering algorithm is displayed by default, but other options are provided to help the user visualise the filter outputs. The menu items discussed below can be found in the Filters submenu of the AVW right-click menu and are only enabled when the filter setting panel is invoked. • Superimpose the raw data. The wave form of the raw data can be superimposed on the filter output. This can be done by checking the right-click menu item [Superimpose Original View]. • View intermediate result. Viewing the intermediate effects of a filter can be useful when adjusting the filter settings in a list of applied filters. To view the intermediate result, check the right-click menu item [View intermediate output]. Under this setting, when a user selects a particular filter from the applied filter list box, the output of that filter (plus the effects of filters applied before that filter) will be shown on the view. 33 3.2.3 Using the filtered data Once the filter or filter combination has been successfully defined and applied to the data, the resultant filtered channels can be added to the variable list and then visualised, searched and manipulated in the same manner as any other channel in the variable list. To add filtered data to the variable list, select any variable to which a filter has been applied and select [Add to List] from the Filters sub-menu of the right-click menu. Alternatively, if multiple filtered variables are selected, select [Add to List] whilst holding down the Ctrl button to add all the selected channels to the variable list. Note that adding the filtered variable to the variable list does not automatically add to, or modify, the contents of the file loaded into SDE. However, the new channels can be saved by selecting [Save Data] from the File Menu. 3.2.4 Applying filters to multiple views Filter combinations applied to one channel can be copied easily to other channels in the AVW. First, apply the filter combination to one of the required channels and ensure this variable is highlighted in red by clicking on it. Then, select the additional variables to be filtered, by clicking on them with the Ctrl button pressed, or hold Shift and click to select a range of consecutive variables. The names of the additional channels selected will be highlighted in blue. Right click anywhere inside the AVW and Select [Apply Filters to Selected Views] from the Filters submenu and the filter will be applied to all the selected views. A common use of the difference filter is to display the difference between a filtered version of a variable and the original data. If such filters are to be copied across multiple variables, a wildcard is required to indicate that each variable should be compared to itself, rather than the variable used to define the filter combination. Replace the text of a variable name with %N% and press Enter to use the wildcard. 3.3 Detect spikes and save spike time data to file The core of SDE is a pattern matching engine; this functionality will be discussed in detail in chapter 4. However, this section will explain how SDE pattern matching technology can be used to detect spikes in the AVW, by using a spike template selected from the feature view of the main view window. To perform spike detection on any channel, first ensure that the appropriate channel is displayed in the feature view window, by selecting it from the control panel. The spike template can then be defined by clicking and dragging over the required area using the middle mouse button. Alternatively, the area can be selected by left clicking on one side of the required section and then right clicking on the other. For more information on selecting templates for pattern matching processes, refer to section 4.1.1. Once a spike template has been selected, switch to the auxiliary view and activate the required channel. From the right click menu, select the menu item [Spike detection]. If one of more matching spikes are found, the detected spikes will show in the same sub-view as vertical lines. However, if no matching spikes are found, the analogue signal will remain. If a shape template 34 is not selected, the right click menu item [Spike detection] is disabled. By changing the subview adjacent to the spike data to the original raw data, it is possible to compare the raw data with the detected spikes, as shown in figure 3.6, with original ECG signal displayed above the detected spikes. The spike time data can be saved to file by clicking on the right click menu item [Save spike time data]. Figure 3.6: Spike dection on an ECG signal Spike detection can be applied to multiple channels simultaneously. To select more than one channel, select the first channel to activate it. Then simply hold down the Ctrl button and click on each additional channel required. The first selected channel will remain active, whilst the additional channel names are highlighted in blue. Alternatively, click on a channel whilst holding down the Shift button to select all channels between this and the active channel. When all channels are selected, click on the [Spike detection] menu item and each channel will be processed in turn. SDE provides a dialogue box, shown in figure 3.7, to enable a user to set the parameters for the spike detection process. The parameter Match threshold is the threshold that defines a valid match of a spike. The maximum matching score is 500, i.e.100 percent match of the template. The lower threshold and upper threshold are the lower and upper boundaries of the value window that defines the amplitude range of a valid matching shape. Given that the amplitude of the template is 1.0, a lower threshold 0.5 and upper threshold 1.5 indicate that the amplitudes of the matches are within 50% to 150% of the amplitude of the template. If the Intelligent template box is unchecked, SDE will simply search for patterns that match the selected spike template. The selection of a typical spike template is therefore very important, because any irregularities in the template will reduce the matching score of any potential spikes. However, by selecting the Intelligent template option, SDE will instead use the supplied template to identify up to 300 matching spikes and take the mean of these to generate a new spike 35 Figure 3.7: Spike detection settings window template. The spike detection is then performed using this new template. In most instances, the spike averaging procedure produces a template that more closely resembles the typical spike shape present in the data and superior results are obtained. If the Save as default box is checked, the input parameters will be saved as the default settings. Each time the parameter setting dialog window is opened, the previous parameter set is used. By clicking the Get Default button, the default parameter set will be loaded and will override the current settings. 3.4 The multiple channel array view MCD data can be displayed in a multi-dimensional array view by selecting [Multiple Channel Array View] from the right click menu of the main SDE window. The multiple channel array view operates in a similar manner to a normal auxiliary view, except that all data channels are automatically displayed in an organised, two dimensional array. It is possible to zoom in and out, play the data, save the view and perform spike detection, but the filtering options are not available. It is recommended that the multiple channel array view is only opened after the whole data set has been extracted from the raw MCD file to prevent the occurrence of “dull” time whilst the full channel set is updated. 36 Chapter 4 Searching for Similar Patterns A key feature of SDE is its ability to search efficiently for patterns similar to regions of interest (“ROI”), in time series data. The user can select or identify a data pattern as the search query; this may be a historical data pattern, an example imported from the current data set, an example imported from another system or a manually created pattern. SDE can then search for this pattern in either the currently opened data or in a precompiled database, potentially containing huge numbers of files. A list of matching results is returned to the user, ranked according to their similarity score, with each providing a reference to the name of the file and the position within the dataset where the similar data section was located. The user can then either save these results as a text file, save the matching patterns as a new datafile or examine them from within SDE. The search engine is based on the AURA (Advanced Uncertain Reasoning Architecture) technology [1], which provides fast, approximate search and match operations on very large, unstructured datasets. SDE can use either a local or a remote search engine, allowing distributed databases to be searched. This chapter provides two scenarios for single pattern searching. Searching for multiple patterns in multiple variables requires the Task Planner, which will be discussed in chapter 5. 1. Searching for a single pattern from any feature in the current data set • Search for similar patterns in the current data set. • Create and manage a ROI template database for the detection of particular conditions. • Multiple query sources - selected ROI, library template or user drawn pattern. 2. Searching for a single pattern from a feature database • Search for the best matches to a query pattern from each feature record in a feature database. • Fast pattern searching of large volumes of data. • Report and sort the similarity scores, record names and positions of the best matches. 37 The searching of databases is initiated via the Search button on the Control Panel in the main view window. To search the currently opened dataset either the Search button or the Pattern Finder can be used. 4.1 Searching the current dataset To search the current dataset using the Search button, the user must first identify a ROI, select the type of search required and define a set of search parameters. 4.1.1 Selecting the ROI Before a search can begin, the user needs to select the example data segment (ROI) that will form the template for the searching process. The ROI can only be selected in the lower (feature view) part of the main view window. Therefore, any feature containing an observed ROI must be first displayed in the feature view window before it can be selected. There are two mouse button combinations that can be used to select a segment of data as the ROI: 1. Left-Right button combination Click the left button at one end of the feature pattern segment and then click the right button at the other end of the segment; the data segment will be highlighted in blue. If the pattern segment is longer than the width of the view window, first click the left button, then go to the control panel click the forward/backward button (or pull the play slide bar) to move the view to the other end of the pattern segment. Finally click the right button to complete the selection. 2. Middle button press/release combination Press the middle button at one end of the pattern segment and, keeping the button pressed, move the cursor to the other end of the pattern segment. Then release the button. If the pattern segment is longer than the width of the view window, keep the middle button pressed and move the cursor from the starting point to outside the view window (in the direction of the end of the segment). The view will “play” until the other end point of the pattern segment is reached. Release the middle button at the appropriate point to complete the selection. There is no limitation on the length of the segment; however, a longer segment usually requires a longer search time. Figure 4.1 shows a selected pattern segment. To cancel the selection, click the left button anywhere inside the feature window. Selecting large sections of data can be awkward within the main view window. Therefore, it is also possible to select a region of interest from the AVW. Hold down the Shift button and use the middle mouse button to highlight a section of data in blue. This data section can then be saved or searched as normal, although large search templates can result in long search times. Press the Escape key to cancel the selection. 38 Figure 4.1: Selecting a Region Of Interest 39 4.1.2 Invoking the search Firstly, to make SDE search the local data, rather than the feature database, ensure that the option [Search within Record], located in the [Preferences/Search Settings] menu, is selected. Next, select your search type from the [Preferences/Methods] menu. This can be a Similarity, Distance or a Threshold based search. Details of these options can be found in section 4.1.3. The remaining parameters that must be defined are the pattern matching Threshold and the Match Low and Match High values, which can be found in the Settings dialogue window, invoked by selecting [Settings] from the [Preferences] menu. The Threshold is a score out of 500, which specifies how similar the shape of the potential match must be to the defined ROI in order to be considered a match, whilst the Match Low and Match High provide the lower and upper boundaries of scale, within which the potential matches must fall to be accepted. More details of these parameters can be found in 1.4. Once all the search parameters have been correctly defined, select the ROI and press the Search button. The search results will be displayed in the Results View Window, sorted by their matching score and the message window will display information about the search. To explore the matched patterns, double click on a selected item in the Results View Window and the main view will move to show the start of the matched pattern. The window will shift so that the beginning of the pattern is placed either at the left-most side of the main view or, if present, at the vertical time-bar (set by middle clicking on the main view). The matched pattern is highlighted in blue. 4.1.3 Search measures SDE provides three measures for pattern matching: similarity, distance and threshold, which can be selected from the menu item [Methods] in the pull-down menu [Preferences]. Details of the different methods can be found in pattern recognition literature, but this section will give a brief overview of each, to assist the user in the selection of the most suitable search measure. Similarity search is based on the correlation between the query pattern and the main signal and is therefore amplitude independent. Two patterns are considered to be maximally similar if they move upwards and downwards at the same time instants and maximally different if the movements are inverted. Consequently, two patterns can contain spikes of greatly differing size and still be considered as a match. Although the initial search is amplitude independent, the user can constrain the search by eliminating potential matches that do not fall within a specified amplitude window. The size of the amplitude window is controlled by the Match Low and Match High parameters, as discussed in section 1.4. The Distance search method is based on the city block distance between each point on the curves and is therefore amplitude dependent. If two patterns contain spikes of greatly differing size, the matching score will be reduced. The Threshold search measure is used to search the data for any areas of activity that lie above or below a threshold level. If threshold is selected, the ROI only defines the length of the query - the shape of the ROI is disregarded. When the Search button is activated, a dialogue window 40 is invoked; asking the user to specify a threshold value and to select whether SDE should search for values greater than, or less than, this threshold. 4.1.4 The search and results view When the search is complete, the returned results are displayed in the results window. There are five columns in the results window. 1. The order of results sorted by the matching score. 2. The name of the record in which the matching pattern was located. When searching the local dataset, this will always be the name of the variable searched. However, when searching a database, this column will indicate the name of the file in which the matching pattern was found. 3. The matching score, - 0 to 500, where 500 represents an exact match. 4. The offset - indicating the position of the similar pattern within the record. 5. The index, within the feature database, of the record in which the matching pattern was found. Figure 4.2 displays a set of search results and shows the menu options produced when right clicking on the result window. These options are: Figure 4.2: The results window right-click menu 41 • Select All. Some operations require the user to select more than one search result at a time. A user can select a set of results by using the conventional Ctrl+Click or Shift+Click methods. However, a user can also use the Select All menu item to select all results items at once. Note that the Ctrl+A key combination does not select all patterns in this instance, to reduce the likelihood of a large number of pattern windows being opened incorrectly. • View Selected Items. When searching a database, search results are displayed in special pattern view windows, rather than using the main SDE data view. View Selected Items opens the pattern view windows for all the search results selected. If more than 8 items are selected, a message window will prompt the user to confirm the operation. • Close Sub-Windows. Closes all the opened pattern view windows. • Copy. Copies the selected item to clipboard. The format of the clipboard data can be found in [2]. • Save. Saves the selected results to a text file. The format of the data is identical to the clipboard data. • Save All. Saves all the results to a text file. The format of the data is identical to the clipboard data. • Save Selected Patterns. Saves the periods of time series data, which correspond to the selected search results, as a new data file in the internal SDE format (or if the file already exists, appends the data to the end of the file). The Save Selected Patterns Window prompts the user to enter values for the left and right span, which control how many data points on each side of the matched results should be saved, along with the matched pattern. If you wish to open the saved data using other tools, it is necessary to remember the length of the pattern (Query length + left span + right span). The query length is reported in the message window. The stored data can also be viewed in SDE. If the check box [Create residual data from selection] is checked, SDE will add another channel to the feature selection list containing the original searched data with the saved results removed. • View Selected Patterns. Opens a pop-up window containing a superimposed view of all the selected matched patterns. The yellow line shows the mean of all the displayed patterns. The view can be saved as a bitmap by selecting [Save View] from the right click menu. 4.1.5 Noise suppression A low pass filter can be applied to the time-series data before searching. This is particularly useful when searching a noisy data set or for a long example segment where the high frequency component is not important. To turn on the filter, check menu item [Apply Filter] on the pull-down menu [Preferences/Search Settings] or uncheck it to disable the filter. 42 4.1.6 Saving feature patterns Feature patterns selected in the bottom right feature view window can be saved to a file or to the clipboard. A right-click menu, shown in figure 4.3, is accessible from the feature view window, containing the items Copy and Save, in addition to the options available in the main view right-click menu. If a segment of a pattern is selected, the right-click menu can be invoked in the normal way, by right-clicking the window. If no pattern segment selection is made, a user can use Shift+Right Click to invoke the menu. Figure 4.3: The feature view window right-click menu 4.1.7 Using the clipboard data Under the Microsoft Window platform, SDE is capable of copying data to the system clipboard. This enables SDE to pass data, via the system clipboard, to other applications. Currently three kinds of data can be copied to the clipboard: the search results, the feature pattern on the feature view window and images of the data view window. The format of the clipboard data can be found in [2] and can be viewed by passing the clipboard data to a suitable open file. The following table summarises the clipboard data used by SDE. Description The Data View Window Search Results Selected pattern interval 4.2 Format CF BITMAP CF TEXT CF TEXT The Pattern Finder In the same way as users of a word processor can search for words in a text file, SDE provides the Pattern Finder to search for patterns within the current data set. The Pattern Finder facility allows the user to maintain and manage a library of pattern templates, with options to import or draw new patterns, recall and edit previously stored patterns and delete templates that are no longer required. The Pattern Finder can then be used to search for one of the 43 templates in the pattern library, within any selected variable of the currently loaded data set. Therefore, by using the Pattern Finder, the user can select a feature from one variable and search for similar patterns in any of the other variables in the currently opened data, or in a completely different data file. The Pattern Finder uses a similarity based search, as discussed in section 4.1.3, with no amplitude restrictions in place (Match Low or Match High) and consequently returns potential matches based entirely on the correlation between the shapes of the query pattern and the main signal. This allows the user defined patterns, created using the draw pad, to be used as a search template. To invoke the pattern finder interface, click on [Pattern Finder] from either the right-click menu or the pull-down menu [Tools]. A screen shot of the Pattern Finder user interface is shown in figure 4.4. Figure 4.4: The Pattern Finder user interface The top window of the Pattern Finder is the pattern view window, used to display the currently active pattern. In the draw pad mode, it is also the drawing area for editing and drawing a new pattern. The lower left corner is a list box containing all the available library patterns and, to its right, there are three source selection buttons used to switch the input resource of the pattern window: Feature Segment imports patterns from the SDE feature view, Library Pattern loads stored data from the library pattern list and Draw Pad allows the user to manually specify a required pattern. A threshold input edit box allows the user to define the threshold for the search. The five buttons on the far right are: • Find - starts a search activity on the currently loaded data set. The feature currently displayed in the feature view window (lower part of the SDE main view window) is searched for the pattern shown in the pattern view window. When the Find button is pressed for the first time, SDE’s main view window will display the first occurrence of a match. 44 Pressing the Find button a second time will move the SDE main view window to display the second occurrence and so on. The SDE message window will display details of each matched location after each button press. • Add - adds the pattern shown in the pattern view window to the pattern library. • Delete - deletes the selected pattern from the pattern library. This action is only valid in the library input mode, i.e. when the Library Pattern button is selected. • Import - imports a pattern from the selected source to the pattern view window. • Close - finishes the searching and closes the Pattern Finder. 4.2.1 Importing a pattern to the pattern view window Select the Feature Segment button and define a ROI in the SDE feature view (the lower part of the SDE main view window). Then click the Import button and the pattern will appear in the pattern view of the pattern finder user interface. 4.2.2 Recalling a pattern from the pattern library Select the Library Pattern button and click on the pattern to be imported from the library pattern list. Press the Import button or double click the pattern name on the library list and the pattern will appear in the pattern view of the pattern finder user interface. 4.2.3 Using the Draw Pad The draw pad mode can be used to edit the imported pattern or draw a new pattern for the search activity. To edit a pattern, it must first be imported to the pattern view of the pattern finder user interface. To use the Draw Pad, whether editing or drawing a new pattern, select the appropriate button to switch to the draw pad input mode. Hold and press the left mouse button and move the mouse to draw or edit a new or existing pattern. Move the mouse with the right button depressed to erase. 4.2.4 Adding a pattern to the pattern library Click the Add button to add the pattern displayed in the current pattern view to the pattern library. Type in a unique name for the pattern in the dialogue box and then click OK. The library pattern list will update automatically. The add pattern operation can be performed in all three input modes. 45 4.2.5 Deleting a pattern from the pattern library To delete a pattern from the library, select the Library Pattern radio button and select the pattern to be deleted from the library pattern list. Then click the Delete button; a confirmation dialogue window will appear. Choose Yes to delete the pattern from the library or Cancel to cancel the operation. 4.2.6 Using the Pattern Finder to search the current data set Perform the following steps to find a pattern in the current data set: • Import a pattern to the pattern view window of the Pattern Finder. This pattern is used as the search query. • Define the threshold for the searching - note that in this instance the threshold is scaled as a percentage (0-100%). • Select the variable to be searched from the SDE control panel, so that the variable is shown in the lower part of the SDE main view window. • Press the Find button. Information about the search will be displayed on the message window. If a matching pattern segment is found, the message will include the position of the similar pattern and the similarity score. In the feature window of the main application, the view is shifted to where the pattern is found. Press the Find button again to find the next similar pattern from the data. If at any time a matching pattern cannot be found, the message window will show “End of record” and the SDE main view window will move to the end of the data set. 4.3 Searching for a single pattern from a feature database While exploring the data, a user may observe an interesting segment or pattern of data. This pattern of data (ROI) may represent an event, fault or an abnormal pattern unfamiliar to the user. The user may wish to search a historical feature database to find out if a similar pattern has occurred previously. If a similar pattern is found, this may provide valuable insight into the situation or suggest possible solutions, by showing how similar cases were resolved previously. By using the Search button, SDE is able to search for a single ROI in a set of historical feature databases. A searching library of the historical data must be established before such a search is possible. Feature databases are created by using the Data Builder, described in chapter 7, which combines any number of specified data files into a single searchable entity. SDE can invoke searches using either the local or a remote search engine. SDE is capable of controlling and displaying parallel searches - more than one search can be invoked and manipulated by SDE in a parallel processing manner. 46 4.3.1 Invoking a database search To invoke a database search, first select a search engine from menu item [Search Engine] from the pull-down menu [Preferences]. If the user is using a remote search engine, a URL needs to be specified in the address window of the parameter bar. Next, to make SDE search the feature database, rather than the local data, ensure that the option [Search within Record], located in the [Preferences/Search Settings] menu, is NOT selected. Select [Settings] from the [Preferences] menu and ensure that the Library Directory field, present in the Settings dialogue window, is set to the location of the database. Select the ROI and search parameters, as with a local search, and then hit the Search button. SDE uses a system of tabs to store multiple search processes, enabling the user to manage multiple parallel searches and to provide easy access to several sets of search results. Once the searching task has been dispatched, a new pane (tab) will appear in the status bar of the SDE window. By default, SDE will store the last ten searches and, when the eleventh is added, will delete the oldest search tab together with any associated search results or dispatched search tasks. To override this, select the pull-down menu [View/Search Results] and uncheck menu item [Auto-delete oldest pane]. With Auto-delete unchecked, once the number of parallel searches is equal to 10, SDE will prompt the user to delete an existing tab to accommodate the new search task. The status of a tab is indicated by its colour: orange - new tab, light grey - inactive, dark grey - active. The Results window displays the searching results of the active tab. The last text message in the Message window displays the relevant information relating to the current results set. The search results and status messages keep updating dynamically until the search is completed or stopped. Figure 4.5 shows SDE with four search tabs available. A search tab is activated by clicking the tab using the left mouse button. To delete the active tab, select the pull-down menu [View/Search Results] and click menu item [Delete Current Results Set]. Alternatively, to delete all tabs at once, select the menu item [Delete All Result Sets]. A search tab can also be removed by right clicking on the tab and selecting Delete. The current search status can be checked by moving the cursor to the tab to view some brief information on the tooltip. The searching status displays the status of the search engine: typically, “dispatched”, “in processing”, “complete”, “cancelled” or “failed”. 4.3.2 Manipulating the search tabs Each of the tabs acts as the interface for that search task with the user. It manages the search results, the updating of the searching task and contains all the information about that search. The name of a tab is the variable name of the feature searched. Using the tabs, a user can switch between the search tasks to view the current status and searching results of the active tab. It also provides controls to terminate the search, save the results and view the full information about the search. Each of the tabs has a right-click menu, as shown in figure 4.6. The menu options are: • View Query Pattern. Displays the query pattern of the search task in a pop up window. 47 Figure 4.5: Search tabs and search results Figure 4.6: The Search tab right-click menu 48 • Search status. Invokes the search status information window, containing a summary of the search details. Figure 4.7 shows an example of a search status window. • Terminate Search. Terminates the dispatched searching task. This menu item is disabled if the search is stopped (completed, cancelled or failed). Otherwise, it is enabled and, if selected, will send a message to the server asking for the termination of the specified searching task. Termination of a searching task does not close the tab and results received so far are still valid for use. Once a search is terminated, its status will be set to “cancelled”. • Save. Saves the current search task, enabling it to be recalled again later. If a remote search using “Pattern Match Controller” (PMC) is invoked that is expected to take a long time, all parameters relating to the current search can be saved and the computer switched off, whilst the search is processed on the remote server. The saved parameters can then be recalled and the results retrieved from the PMC searcher. Note that PMC is not supplied as standard with SDE and users interested in deploying distributed searches on multiple, remote data should contact Cybula for more information. • Delete. Deletes the search tab. By default, it also deletes the search task dispatched unless a Save action has been invoked before the tab is deleted. Figure 4.7: A search status window Some other additional operations relating to the search tabs can be found on the pull-down menu [View/Search Results]: • [Delete Current Result Set] is equivalent to the search tab right-click menu item [Delete]. It deletes the active tab and stops the search task. • [Delete All Result Sets] deletes all tabs and result sets and terminates all the searching tasks. 49 4.3.3 Viewing the search results The search results of each task can be displayed in the result window by activating the relevant search tab. The message window will show the latest searching status of the active tab. If the searching task of an active tab is not complete, the results list on the results window and the current searching status will be updated dynamically. From the results view window, the user can view each result pattern by double clicking on the appropriate row of the results list. Result patterns can also be invoked from the right-click menu of the result window. A pop-up window of a fixed, standard width displays the selected pattern (see figure 4.8). The patterns found during the search activity can be viewed irrespective of whether the feature databases are local or remote. Figure 4.8: The pop-up feature view window The data segment matching the input query is highlighted in red. If the length of the pattern is less than the window width, the rest of the data display is shown in grey. If the length of the matched data segment is larger than the width of window, no data outside the matching segment is displayed. Information relating to the data is also displayed in the window. A scale on the left (or right) side of the window provides a rough measurement of the amplitude of the data. The exact value of the data can be obtained by moving the cursor around the window. At the top of the window, the name, amplitude and offset of the data can be found. Offset is the X-coordinate (in points) of the current position of the cursor, relative to the start point of the original raw data. Amp shows the amplitude of the signal at the current cursor position. Amplitude is not affected by the Y coordinate of the cursor. There is no unit shown for the scale, which is data dependent. Several simple controls are designed to allow the user to adjust the view of the window: • Move the scale. By pressing the space bar, the scale can be moved from the left side to the right side or vice versa. • Move the data pattern. If the length of the data is longer than the width of the window, a user can press the Left or Right key to move the pattern forwards and backwards. • Move the data to the end portion. Pressing the End key will cause the window to display the last part of the data pattern. 50 • Move the data to the beginning. Pressing the Home key will rewind the data view to the start of the data segment. • Change the scaling factor. The scaling factor can be adjusted by pressing the Up or Down keys. • Shift the boundary of the scale. Pressing the Up or Down key, while holding Shift, alters the boundaries of the scale to higher or lower values. This operation does not change the width of the scale. Figure 4.9 shows a set of pop-up result windows. A result window can be closed separately by clicking the Close button on the top right corner of the window. To close all the pop-up result windows at the same time, click the menu item [Close sub-windows] from the right-click menu or use the CTRL+SHIFT+W key combination. Figure 4.9: Pop-up result windows 4.3.4 Variable names In many instances, each variable present in a particular file will correspond to a separate feature and, when searching the historical data, the user will only be interested in finding matching patterns from that feature database. For example, if each file contains ECG and blood pressure readings for a particular patient, the user would want to search the ECG database for patterns that match a section of the ECG data and the blood pressure database for patterns that match sections of the blood pressure data. In such instances, the variable name can be left simply as 51 the feature name and the appropriate database with that name will be searched for matching features. A header of the corresponding .csv file would therefore be of the form: BP,ECG,EEG,Resp However, if the data file contains the same test results but for multiple patients, the simple naming structure is no longer appropriate. The user will need to know to which patient each reading belongs, but may want to search the historical data for matching patterns from any patient, rather than just that individual. SDE therefore needs to be provided with both a variable and a display name, separated by a colon. The text before the colon is considered as the variable name for searching purposes, whilst the text after the colon is used in the feature list, for navigation within SDE. Consequently, the .csv file header will be of the form: BP:BP(Bob),ECG:ECG(Bob), BP:BP(Pete),ECG:ECG(Pete) 52 Chapter 5 The Task Planner The status of many industrial processes is often characterised by more than one variable over time. Consequently, users may need to characterise several parameters over time to describe one condition. Searching for multiple patterns from time-series data is a much more complicated process than searching for a single pattern. The process is not just a simple combination of separate single pattern searching procedures but requires efficient collaboration of all searching procedures, intelligent management of the applied searching constraints and a clear, meaningful output. To fulfil the multiple pattern searching requirements, SDE provides the Task Planner, which allows a user to plan and manage tasks that search for multiple patterns across multiple variables. The Task Planner enables the creation of a task library, with the user able to define a set of task descriptions. Each task description can be considered as a “recipe” for a particular event or condition, which can be applied to constrain the searching when required. Integrated within the Task Planner is an interactive result manipulation window, which provides an interactive user interface to start the search and view the returned results. To open the task planner, select the menu item [Task Planner] from either the pull-down menu [Tools] or the right-click menu. The Task Planner consists of two tabbed views: the task view and the results view. The task view is used to establish and manage the task descriptions and pattern libraries, whilst the results view enables the user to invoke the search activity and visualise the results. The results view automatically tracks the contents and changes made in the task view and provides up to date information to the user on the current searching tasks. The task view or the results view can be selected by clicking on the appropriate tab. Select [Exit] from pull-down menu [File] to close the Task Planner. 5.1 The task view Figure 5.1 shows the Task Planner with the task view tab selected. On the right of the window are the control buttons in three groups. The four buttons in the Task group are used for manipulating the task descriptions. The button groups Library and Sources and the pattern view and library pattern list (both at the bottom of the view) are the equivalent of their counterparts on the pattern finder and are used in a similar manner. Tasks and their descriptions are listed in the main part of the task view. 53 Figure 5.1: Task Planner – task view Each task consists of the task name and several sub-tasks shown as a collapsible tree. For each task there are three parameters listed: these are global parameters and apply to the whole of the task or the final result. Each task is made up of a number of sub-tasks - one sub-task represents the search for a single feature. Each sub-task has a row of parameters, including the name and threshold and an icon indicating the query pattern that will be used. Double clicking on the sub-task icon loads the query pattern into the feature view. The task tree can be sorted by clicking on the head of the Tasks/Patterns column. The functions of the buttons are as follows: • New - create a new task. • Add - add a new sub-task to the task. • Load - load a library task to the tree list box. • Save - save a task in the tree list box to a file. • Insert - add a new pattern to the pattern library. • Delete - delete a pattern from the pattern library. • Import - import a pattern to the feature view window. 54 The three radio buttons (Pattern Segment, Library Pattern and Draw Pad) are used to select the source from which the pattern will be imported to the pattern view window. The task view also has a right-click menu that provides additional operations, which will be discussed in section 5.1.1. 5.1.1 Creating a new task To create a new task, click the New button and a new task will appear in the tree list view window. Double click the new task name and type in an appropriate name for the task. Use the Add button to add a new sub-task to the task, which can also be renamed by double clicking. For each task, there is a set of global parameters that apply to the task. These are as follows: • The task Thres cell defines a threshold for the total score - the final result. If this threshold value is set as a negative value, no global threshold is applied. • The task Tw cell defines a time window width to constrain the search - all sub-task matches must happen within the time interval Tw (seconds). If Tw is set as a negative value, the first sub-task in the task is set to be the principle feature. By setting a principle feature, the search is constrained in such a way that the pattern of the principle feature must happen first and all the other features must appear after the principle feature, in the time interval defined by kT wk, for a match to occur. • The task Condition cell defines the method for the computation of the final result. This can be either the Euclidean Distance or an AND operation. Currently, only the Euclidean Distance measure has been implemented. The parameters for each sub-task are set as follows: • Variable Name - identifies the feature that will be searched and is selected from a drop down list of possible features. Features can be added to the list in the “Signal name selection” dialogue box, activated by selecting [Settings/Variable names]. The variables available for selection will be those defined in the user’s feature def.cfg file (see section 7.1 for details). • Filter - a filter can be selected from the defined filter library. The selected filter will be applied to the data before the search begins. • Measures - the user can select to use either Similarity or Distance as the search measure. • Thres - defines the threshold applied to the new sub-task. Use the Add button to add further sub-tasks to the new task and set the new sub-task parameters accordingly. After the parameters for the both the task and sub-tasks have been set, example patterns are required to be selected for each sub-task. To select an example pattern for a sub-task, the pattern must first be imported to the pattern view, using the same techniques that were used 55 to import patterns into the pattern finder. Then the example pattern can be assigned to the sub-task by clicking the menu item [Set Template] on the right-click menu, whilst the mouse is over the desired sub-task. The sub-task icon will change accordingly to show the pattern. To examine the example pattern assigned to each sub-task, double click the sub-task icon; the example pattern will then be displayed in the pattern view. To delete a task or sub-task, select the task or sub-task and then click the menu item [Delete] from the right-click menu. The right-click menu also provides Cut, Copy and Paste functionality to enable fast copying and movement of tasks and sub-tasks within the task planner. 5.1.2 Saving and loading tasks to/from a library file All tasks displayed in the Task View can be saved to a library file. To save the set of tasks to a library file, press the Save button, define the file name to be saved and click OK. To load a previously saved task library file, click the Load button and the file input dialogue will open. Select the path and file name and then click the OK button. The library file will be loaded and will overwrite the current tasks in the Task View. 5.2 The results view To use a task defined in the task view for searching or finding, the task must be selected from the Task drop down list at the bottom of the view. This lists all task names currently available in the task view and, when the results view is activated, is set to the task selected in the task view. When the user selects a task, the following operations may be performed: • The Search button searches for matching pattern combinations in the various feature databases. • The Find button searches for matching pattern combinations in the current data set. Figure 5.2 shows the Task Planner, opened with the results view visible, after a Search has taken place on the task “Deceleration”. The main body of the results view is the list box, which shows all the search results, with each row representing a separate match. The first column is the record name, which is used to locate the data in the time series data store, and the following columns show the score associated with each sub-task. Consequently, the number of columns in the results view depends on the number of sub-tasks defined in the task carried out. The last column is the total score - a measure computed using the conditions selected in the task description. 5.2.1 Using “Search” to perform a multiple pattern search on historical data A task can be applied to search the feature databases (the historical data) by using the Search button. To perform a multiple pattern search on historical data, first select a task from the drop down list and then click the Search button to start the operation. Once the search has been completed, the results view will be updated and the Applied Task label box will display 56 Figure 5.2: Task Planner – results view the completed task name. The ordering of the results can be changed by clicking the head of a sub-task column to sort the results by the scores in that column, either ascending or descending. The actual matched pattern for each sub-task of each search result can be viewed by double clicking the appropriate score; a pop up window opens to display the pattern. Alternatively, double clicking on the name of a record opens an AVW containing the matched pattern for all the sub-tasks in that task. An example can be seen in figure 5.3. The window displays the record name and sub-task names so that the user can clearly identify the results. The pop up windows can be closed separately by clicking the close button on the top right corner of the window. Alternatively, all sub-task result windows can be closed together, by selecting [Close All Single View Windows] from the right click menu or by using the Ctrl+Shift+W key combination. Selecting [Close All Group Windows] from the right click menu will close all the grouped task result windows. 5.2.2 Using “Find” to perform a multiple pattern search on the current data set A task can be used to search the current data with the Find button. To perform a multiple pattern search on the current data set, first select a task from the drop down list and then click the Find button to start the operation. Once the search has completed, the results view will be updated and the Applied Task label box will display the completed task name. The format of the results is exactly the same as when performing multiple searching tasks on historical data. 57 Figure 5.3: Viewing a match using an AVW However, for results returned from a Find operation, double clicking on a sub-task score does not lead to the opening of a pop up window. Instead, the view on the main application window will interactively shift to the position where the result was found. 5.2.3 Using the right-click menu of the results view A convenient way to explore the search results from the task-planner results view is using the right-click menu. Figure 5.4 shows the right-click menu of the task planner results view. The actions of each menu item are as follows: • Item Details - displays the details of the selected search record. • View Group - displays the patterns matched for all sub-tasks of a search record in a single, decoupled, auxiliary window. Users can play, zoom in/out and scale the view in the auxiliary window to examine the details of the signal and can save the view as a bitmap image. • Close All Single View Windows - closes all the separate sub-task, result, pop-up windows. • Close All Group Windows - closes all the grouped task result, pop-up windows. • Invoke Tool - invokes a post-processing tool to analyse the search results further. This operation has not yet been implemented. 58 Figure 5.4: The right-click menu of a task-planner results view 59 Chapter 6 The SDE Protocol A pseudo protocol called SDEP is designed to pass commands and parameters from a web page to SDE. The SDEP protocol starts with tag sdep:// and follows with action tag, source data URL and other options. The general format of a SDEP protocol string is: sdep://action|data-source|option-1|option-2|...|option-n| In the string, action and data-source are the basic parameter fields. Currently, the available actions controlled by the SDE protocol are start, open, search and close. Depending on the actions, one or more options can be appended to a protocol string. The standard format of the option string is: P arameter name = value string Fields of the protocol string are separated by the character ‘|’. The end ‘|’ of a string is optional. 6.1 Actions and data sources The action field defines the actions that SDE should perform. If the action requires data from the remote site, the data-source field must be used to provide a URL for SDE to download data from the source. 6.1.1 The action field The action string can be one of Start, Open, Search and Close. • Start - starts one instance of SDE. If SDE has already been invoked, this action will be omitted. This action does not require any parameters and any following parameters will be ignored. • Open - tells SDE to open a data set, defined by the URL, in the data source field. If SDE has not been invoked, this action will execute the Start action before opening the source data. Options to define a specific data set and the display time interval can be defined 60 in the option fields. The order of the options is not important. The Open action is the default action of the SDE protocol and, consequently, the action field of an open action can be omitted. • Search - asks SDE to search for patterns that are similar to a query pattern. The query pattern is defined in the data file that is located at the URL supplied in the data source field. The format of the query pattern data file can be either CSV or TDD data format. All other parameters for the search, such as search engine, metric, threshold, etc. can be defined in the option fields. • Close - closes SDE. No parameters are required for this action. If SDE is not open, this action has no effect. 6.1.2 The data source This field defines the URL from which SDE downloads the data files required by the defined action. SDE supports data downloading from HTTP, FTP, SRB and the local site. Unless the official format of the SRB URL string is published, SDE will use srb:// to present the SRB data URL. 6.1.3 The option field The option strings are operation dependent. Details of the option strings can be found in [2]. 6.2 Examples of using the SDE protocol This section provides a series of examples that demonstrate how the SDE protocol can be used to control SDE from a web page. These commands can be tested by typing each SDE protocol string into the address bar of a web browser and then pressing Return or clicking the Go button besides the address bar. 6.2.1 Starting SDE To start SDE the SDE protocol is: sdep://start 6.2.2 Closing SDE To close SDE the SDE protocol is: sdep://close 61 6.2.3 Opening a remote data file The following string will open a remote data file and display the data from point offset 1000. sdep://open|ftp://ftp.cs.york.ac.uk/pub/bojian/testfile3.tdd| point offset=1000 6.2.4 Searching for a pattern To perform a search, it is first necessary to make a data file for the query pattern. This section will explain how a query file can be manually created. The query is actually just a TDD or CSV format data file containing the relevant short period of time series data. To make a single pattern query file, Open a data file in SDE and add a subset of the available features to the main view window using the right-click menu. Make sure all the features displayed in the main window are supported by the searching library. Play the data and find an interesting pattern. On the bottom feature view window, select the time interval of that pattern segment and click on [Save Data] in the pull down menu [File]. In the file save dialogue window, select the output file type, (for example .csv), type the output file name and press the Save button. The SDE protocol can now be used to invoke a search from a web page. • Invoke a search for the local file D:/testing/query1.tdd using the local searching engine and apply noise depression to the data. sdep://search|file://D:/testing/query1.tdd|search engine=local |filter=-1 As the default search engine is local, the above string is equivalent to: sdep://search|file://D:/testing/query1.tdd|filter=-1 • Invoke search for a remote file using the default PMC searching engine and perform noise depression on the data. Use the data indexed 0, 7, 8 and 9 in the file as query patterns to invoke multiple searches. sdep://search—http://www.cs.york.ac.uk/ bojian/query8.tdd| search engine=PMC|indexes=0,7,8,9 The option search URL: explicitly defines the address of the master PMC node sdep://search|http://www.cs.york.ac.uk/ bojian/query8.tdd |search engine=PMC|indexes=0,7,8,9|search URL=http://pascali. wrg.york.ac.uk:8080/ogsa/services/PMC/PMCControlService 62 Chapter 7 The Data Builder Although SDE can be extremely useful to visualise and explore individual datasets, much of its most powerful functionality is based upon searching for important or unusual patterns in multiple, large databases of precompiled time series data. This chapter will discuss how these databases can be compiled and incorporated into SDE. There are two stages in the data building process. Firstly, the variables that will be stored must be defined in a variable configuration file and then the data libraries are built using either the data builder command line interface or the Build Search Library tool inside SDE. 7.1 The variable configuration file Before a historic feature database can be built or searched, the variables that will be stored must be defined in a configuration file. The purpose of the configuration files is to provide a link between each variable and the name of the corresponding database and to control the display of each variable in SDE. Both SDE and the data builder need access to a configuration file in order to access or modify the database contents. The SDE installation directory contains a template configuration file called “feature def.cfg”, which is copied to the location specified below in each users file system, when that user first starts SDE. Note that this location is often hidden from view in Windows Explorer, but this can be altered by selecting Show hidden files and folders from the View tab of the Tools→Folder Options dialogue window. C:/Documents and Settings/USERNAME/Local Settings /Application Data/SDExplorer SDE will use the variable definitions contained in the feature def.cfg file in this directory when displaying and searching data. Consequently, different users on the same machine can use SDE with a different set of variables defined in their individual configuration files. The data builder can be linked via the command line to the feature def.cfg file in the installation directory, the local directory or to any other valid configuration file. 63 The first line of the configuration file is used by SDE to determine the initial location of the search database. It can be altered from inside SDE by accessing the Settings page in the Preferences menu. The default search directory has no effect on the data building process. Following this initial line, a separate record needs to be added to the configuration file for each variable, containing the variable name, the database name, the range of values it can take, a quantisation step and a unique ID number. Note that multiple variable names can be linked to a single database name, in which case a search for a ROI in any one of the variables will potentially match features in any of the linked variables. No spaces are permitted in the database name. If you are unsure of what quantisation step to use, set it to 0.00 and the data builder will select an appropriate value. Likewise, if both the upper and lower bound are set to 0, SDE will automatically control the display range. The format of a variable record is: UpperBound LowerBound AerospaceVibrationDataOnly AerospaceVibrationDataOnly QuantisationStep DatabaseName IDNumber VariableName Therefore, the entry below will define a variable ‘ABC’ that has a value range of 0 to 1000 and associate it with a database called ‘LibABC.dbs’. The unique ID number for this variable is 601 and the quantisation step of 0.00 instructs SDE to select a suitable value. For any data other than aerospace vibration data, set lines 2 and 3 to ‘1 0 0 0’ and ‘7’ respectively. 1000 0 1000 7 0.00 LibABC.dbs 601 ABC 7.2 Building a feature database using the command line interface Once a configuration file has been created, the data builder command can be used to add any of the defined variables that are contained in a CSV file to a feature database. The data builder is available in both Windows and Linux and both function in an identical manner. However, when building very large databases, it is worth considering that the Linux version is significantly faster. Note that the data builder command is currently only able to create databases from CSV files - neuroscience data must be built using the internal Build Searching Library tool, discussed in section 7.3. 64 In order to add data from a CSV file to a database, using the example variable configuration defined above, one of the features in the CSV file must have the variable name ABC. Note that if both a variable and a display name have been defined in the file header, the variable name is the text before the colon (see section 4.3.4 for details). SDE will convert the data in this column to a DM vector and append the vector to the file LIBABC.dbs. If the other columns in the raw data file are also defined in the configuration file, they will also be converted and added to the relevant .dbs file. To run the data builder from the command line, first navigate into the Databuilder directory, which, if located in the default position, can be done by typing: cd c:/Program Files/SDExplorer/Databuilder The databuilder command can then be used to compile the data, according to the information provided using the flags listed below. -b -m -v -p library location source file or directory of source files configuration file file name tail size and replacement string The library location is the directory in which the database should be built. Note that multiple databases can be built in different directories, using the same configuration file, and SDE can select which should be searched by altering the Library Directory field in the Preferences/Settings dialogue window. The -m flag is followed by either the location of a .csv file that is to be added to the database, or a directory, from which all the CSV files in that directory should be added to the database. The -p flag is followed by two parameters: the file name tail size and the replacement string. The file name tail size is the number of character that must be removed from the end of the source files to produce the record name reported during searching. These are then replaced by the replacement string. For example, to remove .csv from the file names and leave just the main data names, the sequence -p 4 “” can be used, whereas -p 4 .csv will remove any three letter suffix and replace it with .csv. The following command will add any variable defined in the configuration file located in the SDExplorer directory, which are present in the file dataFile1.csv, to the database located in c:/dmvlib: Databuilder -b c:/dmvlib -m “c:/documents and settings/ USERNAME/my documents/new data/dataFile1.csv” -v ../feature def.cfg -p 4 .csv Alternatively, the following command will add all of the .csv files present in the “new data” directory to the database: 65 Databuilder -b c:/dmvlib -m “c:/documents and settings/ USERNAME/my documents/new data/” -v ../feature def.cfg -p 4 .csv When building the database in Windows, the library location may not contain any spaces. Spaces are permitted in the source file location, or when using Linux, but inverted commas must be used to ensure that the command is parsed correctly. It is also important to note that as the data builder appends the supplied data to the existing feature database, if the same file is added to the database twice, any matched patterns contained within the file will be returned twice. If you wish to replace the existing database with the new data, it must be manually deleted before the data builder is utilised. 7.3 Building a feature database from inside SDE A library can be built using either CSV or neuroscience data from within SDE, although a suitably defined configuration file is still required. The builder tool can be invoked by selecting Build Searching Library from the Tools menu and a screenshot of the parameter input window is shown in figure 7.1. Figure 7.1: Building a searching library from inside SDE Four parameters are required: • The Input Path is the location of the data that is to be built. This can be either an individual file or a starting directory for auto seeking input (see below). • The Output Directory is the location in which the library will be built. • The Configuration File must be set to a valid variable configuration file, defining the names and build parameters of the variables that are to be added to the database. • When Auto seeking into input directory is deselected only the selected input file is added to the search library. However, if auto seeking is enabled, SDE will recursively add 66 all valid input files contained in the selected directory and any sub-directories to the search library. To commence the building process, press the OK button. The SDE window will be hidden during the building process and will reappear upon completion. Note that building large databases when using the Windows operating systems can take a long time (approximately 5 minutes per gigabyte). 67 Appendix A The SDE Licence Manager SDE is supplied with a free 30 day trial, during which all the functionality of SDE is available to the user. After this period, SDE can no longer be used to open, view or search data files until either the Standard or Professional version is purchased from Cybula Ltd. The Standard version of SDE contains all of the visualisation, dynamic filtering and spike detection functionality currently present in SDE and allows basic single feature searching. However, the Task Planner is disabled. The Task Planner is only available as part of the Professional version of SDE, which allows the user to create search tasks that identify multiple coincident patterns across multiple features. Further information on both products can be found at www.signaldataexplorer.com or by emailing [email protected]. If an annual licence is purchased, both the Standard and Professional versions of SDE will last for a period of one year, and must then be renewed. Alternatively, in some circumstances a permanent licence can be purchased, which does not expire. SDE contains a licence manager interface, which provides information regarding the current state of your licence and the functionality required to purchase, upgrade or deactivate your licence. To invoke the licence manager interface, select [Registration] from the [Help] menu. If less than 7 days of your licence period remain, or if your registration has already expired, the licence manager interface will be shown automatically when SDE is started. An image of the SDE licence manager interface can be seen in figure A.1. The Licence Status indicates whether SDE is currently running in Standard, Professional or Trial mode and the Expiration Date shows the date and time at which the current licence will terminate. To return to SDE, click the Continue button. Note that if your SDE licence has terminated, the Continue button will not be available until your licence period has been extended. A.1 Purchasing SDE Standard or Professional Edition SDE purchases are performed using a code exchange. Firstly, the licence manager interface is used to generate a request code for the version of SDE you wish to purchase. This request code should be sent to Cybula, along with your contact details, using the web registration form or via email. A Cybula representative will then contact you to arrange payment and provide you 68 Figure A.1: The SDE Licence Manager with an activation code, which will unlock the appropriate functionality and extend the licence period. Note that the activation code will only work on the machine from which the request code was generated. To generate a request code, click on the Buy SDE Standard or Buy SDE Professional button on the licence manager interface. A dialogue window will be invoked, containing the Authorisation Request Code and three buttons: Web Registration, Input Activation Code and Cancel. The Web Registration button will link you directly to the Licence Management section of the SDE website. Enter your contact details, click the Submit button and a Cybula representative will contact you shortly. Alternatively, the request code and contact details can be emailed to [email protected]. Once you have received your activation code, click on Input Activation Code, then enter the code and press OK. Your licence will be updated. 69 A.2 Upgrading or Renewing your Licence Once SDE has been purchased, the Buy SDE Standard and Buy SDE Professional buttons will not available until less than 30 days of the licence period remain. The exception is that the Buy SDE Professional button will remain when SDE is running in standard mode. This allows the user to upgrade from Standard to Professional for their remaining licence period. When less than 30 days of the licence period remain, the Buy SDE Standard and Buy SDE Professional buttons are reactivated, allowing SDE to be renewed and a new activation code purchased for an additional year of use. When the licence is renewed, any remaining days from the previous licence period are added onto the new expiration date. A.3 Migrating your licence to a new machine SDE is sold on a per-machine basis, with a separate SDE licence required for each machine on which it is to be run. However, if a machine is upgraded or replaced, it is possible to migrate the remaining licence period from the existing machine to the new machine. To migrate a licence, first click on the Deactivate Licence button on the SDE licence manager of the machine that contains a currently active licence. This will prevent any further use of SDE on that machine, but will issue the user with a deactivation code. If you send both this deactivation code and the request code from the new machine to Cybula, an activation code for the new machine will be issued for the remaining licence period, at no additional charge. To recover a machine that has been deactivated, so that SDE can be run once again, contact Cybula at [email protected]. Note that this will require a fresh purchase of SDE or the migration of an existing licence period from another machine. A.4 Extended Trials, Multiple Licence Discounts and Permanent Licences To discuss the possibility of extended trials, multiple licence discounts or non-expiring licences, contact Cybula at [email protected]. A.5 Updating SDE Whilst your SDE licence remains valid, you can update to the latest version of SDE at any time. The latest version of SDE can be downloaded from the website below. When the older version of SDE is replaced, your licence status and expiration date are retained. http://www.cybula.com/signaldataexplorer/download.htm 70 Appendix B Shortcut Keys Main view window Ctrl + C Ctrl + E Ctrl + F Ctrl + O Ctrl + P Ctrl + R Ctrl + T Ctrl + Shift + D Ctrl + Shift + W Alt + N Copy Clear message window Start pattern finder Open dataset Print Open remote file Start task planner Delete address list Close all sub-windows Open auxiliary view window L N T U Moves the view back to selected data segment Displays the next feature from the feature list in the feature view window Displays the previous feature from the feature list in the feature view window Switch between point and time offset Recall default view settings Up Down Shift + Up Shift + Down Left Right Increase y-scale Decrease y-scale Increase y-offset Decrease y-offset Change variable Change variable P 71 Space Page Down Page Up Change scale position Activate the next variable in the feature view window Activate the previous variable in the feature view window Auxiliary View Window Ctrl + A Ctrl + C Select all Copy A F G I L M Q R S V Y Ctrl + Y Apply filters to selected views Reset scaling factor Apply scaling factor to a group of superimposed views Adjust vertical scale of selected view according to currently visible data Moves the view back to selected data segment Switch view mode Quick settings panel Remove filters from selected sub-views Apply scaling factor to selected views Superimpose selected views Add just the selected view to variable list Add all highlighted views to variable list Up Down Shift + Up Shift + Down Left Right Increase y-scale Decrease y-scale Increase y-offset Decrease y-offset Change variable Change variable Page Down Page Up Activate next variable Activate previous variable 72 Results Window Space End Home Change scale position Move to end of data Move to start of data Save data and channel selection dialogue window Ensure the right side list box is activated Ctrl Ctrl Ctrl Ctrl + + + + A U S R Check all items Uncheck all items Check all selected items Uncheck all selected items 73 Bibliography [1] J. Austin. Distributed associative memories for high speed symbolic reasoning. International Journal on Fuzzy Sets and Systems, 82:223–233, 1995. [2] B. Liang. Data structures used by the sde for control and data exchange purposes. BROADEN document, BROADEN, York, TN06.002 Version 1b., 2002. 74