 |
|
|
 |
What's New?
Version 4.0 is well underway and will provide our user community with a number of advanced features and capabilities to better analyze data. As with any new version, the performance, appearance, and throughput have been greatly improved. We have also put significant effort into ease-of-use issues and have reorganized many of the menus and interfaces to better accommodate a more refined analytical flow. We have already started to beta-test the version with select partners and integrators and are incorporating their feedback.
Over the next several newsletters, we will start to introduce our readers to the new 4.0 features. Not only will the much anticipated LAS-NVG coupling be available (additional charge applies for this license); 4.0 will also support some basic text-mining capabilities including thematic categorization, classification, and entity extraction. All text processing is based on regular-expression matching, which has proven to perform considerably well in the diverse domains supported by VisuaLinks. Although shipped with basic defaults, the text-mining configurations are customizable by the end user or administrator and can therefore be tailored to address very specific scenarios.
4.0 introduces better designs in the placement settings, menu layouts, licensing manager, and presentation control for objects and associations (colors, thickness, labels, etc). The VBase subsystem has also been updated with the latest code to provide more reliable and manageable control over the underlying data and parameter storage. Additionally, in the 4.0 version, the server-side APIs have been implemented allowing third-party developers to control any of the server-side services.
Look for a more in-depth discussion in the next newsletter as we start to describe these changes and provide our readers with examples.
|
|
|
 |
 |
...that the Group Selection can be turned off when clicking in the area of a cluster?
Some users like this feature - some find it distracting. As with many VisuaLinks features, you can turn this option on or off, depending on your preference.
You can set your Group Selection preferences from the Main Settings window. Use the Edit menu's Main Settings command (in the View area menus) or the Main Settings icon in the Bottom Toolbar to access the Main Settings window.
On the Main Settings window, switch to the General settings.
In the middle of the window, you will see an option called Group Selection that
is selected by default. Simply click this option to deactivate the feature.
When the Group Selection option is turned on, clicking anywhere inside a cluster
will automatically select the entire cluster as shown below.
With the Group Selection option turned off, you can click anywhere inside a cluster without activating it - and can also drag selection boxes to more easily select individual objects within the cluster as shown below.
Note: Group Selection provides a quick way to move clusters of data around the display - used most often when viewing the data from a high level (full fit). Sometimes, zooming in on the display and trying to grab an area will activate the Group Selection and select all objects rather than the desired entities. Determine what setting works best for your analytical needs.
|
|
|
 |
We have been talking about the VAI-LAS coupling since the beginning of this year and soon,
our user community will be able to get their hands on this very powerful capability.
Through our strategic partnership, VAI and LAS have combined technologies to enhance the
ability to find name variations in databases. Specifically, the Name Variation Generator (NVG)
offered by LAS was incorporated into a variety of the analytical services offered within
VisuaLinks including Query, Name Matcher, Duplicate Detector, and Disambiguator.
For this example, we are showing how it applies to the Alias feature contained within the
Advanced Query interface. In the commonresources directory there is a file called Alias.txt
that is used to seed the system with alternative spellings of common names. This file can
be used to "alias" anything including ZIP codes to congressional districts, slang names for
different drugs, or abbreviations for corporate names. The illustration below shows an example
of what the interface looks like for constructing a query.
Once the model, type, and attribute are defined, the value field can be referenced for aliases by clicking the
icon  that looks like an index card. At this point, if there are any matches for the value in the Alias.txt file,
they will display in a list. In this example, the list displays the 7 matches that were found in the Alias.txt
file for the value we entered (MOHAMED).
Additional values can be generated based on the names of people using the LAS-NVG component.
The LAS-NVG uses culturally-sensitive rules drawing from nearly one billion names derived from
more than 200 countries to produce its results. To learn more about the NVG, please visit the
LAS web site (www.las-inc.com)
The  button provides access to a set of
parameters that can be adjusted to determine how "close" the name variation is to the value
presented in the query. The following window appears to provide control over the culture and
the weight of the match.
By asking for all values (not biasing any of the parameters), the LAS-NVG produces a
considerable number of very good alternative spellings of the name MOHAMED. In fact,
almost 100 different spellings of MOHAMED were generated by the LAS-NVG and the examples
shown below represent just a small sample of the final list.
When the query is submitted by VisuaLinks, all selected aliases will be presented to the
database and any matches will be returned. This capability has proven to be quite effective
in a variety of domains looking for name variations.
For additional information on other VisuaLinks components integrated with the LAS code, please review the
Integration of LAS and VAI white paper
available on the Visual Analytics web site.
|
|
 |
VAI supports many of our clients with modeling their data sources, analyzing their contents, and reporting on the results. This usually involves integrating multiple databases through VisuaLinks. Depending on the nature of the data, the sources often contain the names of people, addresses, phone numbers, IDs, or accounts.
This month's link chart illustrates the ability to integrate data from multiple sources using the sample data/models shipped with VisuaLinks. These sample sources have been "tweaked" for this demonstration.
The first source utilized is CRIMINAL-INVESTIGATION and the subject with the name Brad Billings is presented in the display.
Showing only the first level of connections, we see that our subject has connections of a variety of different objects including a criminal organization.
At this point, we use all the information pertaining to our criminal subject and cross-check the MONEY-LAUNDERING model to see if there are any reports filed on him.
Immediately, we see that through a shared driver's license number, our subject (identified with a different date-of-birth) is connected to three SAR transactions (suspicious activity reports) that all link to a common account number.
From here, we check the WIRE-TRANSFER dataset to see if any money has been moved in or out of the account.
There are 10 deposits, all below $10,000, that appear against this account. Someone is transferring money into the account and most likely our subject is moving it out in cash to support the criminal operations associated with his affiliated gang.
Using a telephone toll database, the phone numbers are checked to see if they have any communication with other parties.
Our subject heavily calls a phone number registered to a different person. That number, in turn, contacts additional phones and, at the third level of connection, our calling network stops. The thickness of the linkages indicates the frequency of the calls made between the phones.
Finally, we check our 5th and final data source to determine if there are any vehicle registrations affiliated within the network.
As can be seen in the diagram, there are a total of 3 people connected to these phones and, in fact,
one of the phones has two connections. The addresses are in a different state from our original subject
and the vehicles can be further checked against outstanding tickets, moving violations, or other criminal activities.
In this example, a total of 5 data sources were cross referenced using the Data Walk feature contained
within VisuaLinks. The common objects (type/key) appeared in each of the sources with their respective associations.
This scenario can also be accessed from the White Papers area
of our web site. Additionally, you can use our Link Charts area
to view the Link Charts presented in all of our previous newsletters.
|
|
|
 |
|
