May 2004  Volume 3, Issue 5    What's New? The training for VisuaLinks 3.0 has been revamped and structured into 3 separate classes (Basic, Advanced, Modeling) to accommodate the different range of capabilities and skills represented in our user community. The redesign provides a better flow for gaining the necessary skills to be proficient at performing proactive analytics and producing better results than can be achieved by competitive products. Basic training is designed as a 2 day class for people without prior exposure to VisuaLinks. The class covers an introduction to analytical methods and provides a thorough overview of the fundamental services within VisuaLinks. In the basic class, students are taught the following: Screen Navigation Query (basic) Query (advanced) Walk Data Details (drill down) Presentation Mode Save Sessions Advanced training provides instruction on the more comprehensive and complex services within VisuaLinks. In this class, proactive analytical techniques are taught in the context of the services and tools available in the advanced settings of VisuaLinks. During this class, students are instructed on the following: Network Miner Summarize Merge Objects Terminate Query Charts/Graphs Temporal Grids/Timelines Parallel Coordinates Batch Query Modeling is geared towards those users that will actually create the mappings between VisuaLinks and the database. This class covers a number of methodological approaches to data modeling and provides hands-on scenarios using a number of different data sets. Although this class is geared more towards technical users, it can be taken by anyone with a fundamental understanding of data and databases. In the Modeling class, students learn the following: Creating/Editing Models Defining Objects/Associations Assigning of Attributes/Keys Database Drivers & Settings Lookup Tables Disambiguator Settings Visual Analytics is preparing a schedule of training classes. In the meantime, if you are interested in signing up and taking our new training classes, please use the Training Request Form on the VAI web site to submit your request. We will let you know what dates are currently available. Here are a few tips and tricks to help you use VisuaLinks quickly and more efficiently. …marking objects can be customized by the user? When using the Mark Objects mode, there is a standard template of icons that can be applied to any selected object. The default setting is the left-pointing yellow arrow. However, at the very top of the options box is an option called "Add Mark Icon" that expands into submenus allowing you to select additional icons for use in marking. The "Show Mark Box" option displays red shading in the background of any marked objects. In the diagram below, the marked object on the right side of the network is emphasized by the red-fill, in addition to the marked arrow icon. The "Scale Marked Icons" option is very useful when looking at the big picture. When there is a lot of data present in a diagram, it is much easier to find marked objects if the icon stands out from the rest of the display. The example below shows the usefulness of this option. The first update to the VisuaLinks 3.0 release has been made available on the VAI Support Site. This update incorporates a number of changes implemented to enhance and strengthen the capabilities of VisuaLinks 3.0. Feature Modification Object Edit The Add/Edit Object features now: - Creates "real keys," based on the model definition, when adding an object to data derived from a model. These keys are created for you, as "string" data. - Allows users to edit the *KEY and real key values Saved Information Sets All services can be run and return appropriate results on Information Sets (saved to the VBase). Network Miner When loading the results of a Network Miner request, users can now select the source from which the values should be loaded. This allows users to load values from a saved Information Set. Verity Text Search The Verity Text Search capability has been enhanced to: - Run in the Basic Database Query mode. - Support queries on multiple text search conditions. - Retain text search settings when a query is saved. Cross Database Query The Cross Database Query service has been modified to ensure that the *COUNT MATCHES column in results display the correct number of matches regardless of how the keys are transformed within the different models being analyzed. Summarize Connection pooling during a load of Summarize service results has been modified to remove the need for multiple connections. List Settings The Lists function has been enhanced to: - Retain use settings (Use All User Settings, Use Current User Settings, or Use Both) between sessions. - Add Hit List items to the Current User's List only. Sessions Sessions properly restore data containing grouped objects. DIG Driver VisuaLinks now allows you to view the contents of documents returned from a DIG text source. Data Tables Special content (remarks, pictures, URLs) in data table columns (such as in the Details or Query Results panels) moves with the appropriate column when columns are rearranged. Time Line Placement The Time Line placement algorithm has been modified to display a separate time line to represent "null" values. Starburst Placement A property has been added to the ./clientresources/ClientResources.properties file allowing users to adjust the number of levels VisuaLinks will consider when rendering large networks with numerous cross-connections in the Starburst placement. The property (com.vai.vl.position.StarburstP.depth) is set to 1000 by default, which is a sufficient setting for most data sets. Increasing this value can suppress a stack overflow error and can more accurately display the clusters in very large data sets. User Password New passwords are now set correctly when using the Change Password command from the User Settings menu. Applet A port property has been added to the Applet and Tips configuration properties to allow administrators to point to a port other than the default (80). Refresh Associations The Refresh Associations command now pulls the correct information from the data source. Classification Settings The Classification settings feature has been enhanced to: - Retain settings when set on an added association. - Save and retain settings adjusted when running the VisuaLinks Client from the Applet. Association Bends Associations have been modified to: - Apply bends to only the selected association when multiple association lines cross. - Proportionately move bends applied to association lines within a group. Custom Disambiguator Functions The following property was added to the ./serverresources/ServerResources.properties file to allow a custom function developer to specify third-party jar files on the compile classpath: com.vai.vl.io.server.AgentServerImpl.compileJavaSource.classpath =./serverresources/drivers/jconn2d.jar;./serverresources/drivers/ojdbc14.jar Derived Types The Derived Types function allows modelers to automatically create objects based on attribute values. This function currently works with the following services: - Database Query - Load - Word Count - Name Matcher - Duplicate Detector - Network Miner - Database Walk Note that Derived Types may not produce the proper results using the Cross Database Query, Summarize, Scheduler and Alerts services. Additionally, data derived from a model using Derived Types should not be saved as an Information Set (to the VBase). This month's Link Chart continues last month's discussion on exposing different types of situations that potentially indicate corporate fraud. One of the more power features of VisuaLinks is its ability to provide high-level counts and abstractions of the underlying data using the Summarize service. Similar to traditional OLAP (On-Line Analytical Processing), Summarize is one of the quickest ways within VisuaLinks to get a breakdown of various data fields such as payments, amounts, dates, and other content where multiple (repeating) instances of the values is considered questionable behavior. Thus, Summarize is often used for understanding transactional behaviors such as payment and invoicing frauds. (Hint: Summarize should be used when a question starts with "How many…") In the following diagram, the results of a Summarize request show the top payments, in terms of frequency, in the system based on the vendor master file. The *COUNT column represents the total number of payments followed by the *GROUPED BY AMOUNT and *GROUPED BY NAME columns. Each row in the Summarize results presents the total number of payments made to a specific vendor for a specific dollar amount. The *GROUPED BY NAME column containing the payee information is intentionally hidden by a gray box to conceal the names. The blue arrows depict the same vendor, which in this case, represents the name of an individual. What is immediately revealed is that this person has been paid 64 times for the exact amount of $96.15 (shown by the top arrow). The nature of her business is unclear; however, basic breakdowns for monthly or weekly reimbursements do not correlate to this type of known payment frequency (e.g., monthly parking, internet fees, mileage reimbursement, lease or rental costs, etc.). Additionally, the second arrow shows 18 payments of $192.30 - which is exactly double the $96.15 payment amount ($96.15 x 2 = $192.30). This can be considered an additional 36 payments of $96.30 which conceptually brings our total up to exactly 100 payments of $96.30. Furthermore, there are no other payment amounts for this vendor in the data, only the amounts previously described. Summarize supports the drill-down of any of the data returned and subsequent review was conducted. The next diagram takes the results shown previously with the *GROUPED BY AMOUNT column resorted to show the highest payment amounts. The blue arrows show a particular payee (a corporation) receiving some of the largest payment amounts recorded. The concern here is to determine if the payments are part of a financing plan (e.g., equipment, construction) or if the payments potentially represent duplicate payments for the same invoices. The red arrows show the payments for a different company where the amounts in question tend to be more "rounded" (whole numbers - no cents) and "clustered" around the same range. In this case, there are 12 payments clustered around the $30,000 range. Additional review of both payment scenarios was initiated to determine the nature of these payments. The final diagram shows the results of a Summarize request where the total amount for all the invoices were summed up for each vendor. The results on the left are sorted based on the count of the number of invoices processed and the results on the right are sorted by the total value of all invoices paid. This diagram has been annotated with lines and arrows to show where the same vendor appears in both results. One interesting observation is that only 6 of the most frequently paid vendors (left) also account for the most cost (right) to the company. Additional functions could have been applied to these results to also show the MIN and MAX values for any particular payment.

To unsubscribe from this Newsletter, use the unsubscribe option in your support site profile. If you do not have a profile, please e-mail support@visualanalytics.com to unsubscribe.