|
|
To Catch A Thief |
| Page 1 >> |
ABSTRACT
To catch a thief, spy, money launderer, insider trader, tax evader, narcotics trafficker or anyone exhibiting patterns of "non-compliant" behavior can be effectively performed using a combination of strategic (proactive) and tactical (reactive) analytical techniques. The exact definition of a non-compliant pattern will vary according to the activities occurring within the domain for which it is being applied. In many instances, the methods used to identify a pattern will change depending on the situation-the challenge to the investigator is to discover what associations in the data constitute a reliable and valid pattern. This paper overviews several methodologies used to identify patterns of non-compliant behavior across multiple and disparate data sets (e.g., financial crimes). Real world examples utilizing an operational visual data mining application targeted at money laundering operations are presented to show exactly what conditions were used to identify different types of behaviors based on a set of transactional events (e.g., the activities).
|
Introduction
In relational data-management systems, or any electronic media for that matter,
data is usually taken at a face value, that is, without trying to understand much about how and what is being represented. We typically gather information and store it into buckets (e.g., database tables and fields) so we can recall it at some future point in hopes that it may be useful or provide value to an ongoing case or investigation.
In the law enforcement community, there are enormous volumes of information that are being collected and stored in this fashion without regard to how it will ever be used. Thus, there are huge repositories of information that exist and most likely may never be effectively utilized. This is due in part because there are no established means (technologies or methodologies) by which to easily understand what is contained in them. Thankfully initiatives supported through government sponsorships and commercial initiatives have provided new approaches to help law enforcement organizations better understand, manage, and present their data. Furthermore, these advances are applicable across a wide range of functional areas.
|
Understanding Data
|
The goal of many law enforcement investigations is to find associations or establish relationships among the targets or suspect entities. Associations can be made not only based on the name of a person or organization but also through addresses, identification numbers (including licenses, passport, or social security numbers), telephones, vehicles, accounts, and most importantly activities. Activities represent the phone calls, border crossings, cash deposits, and meeting contacts associated with the illicit conduct being analyzed. A single occurrence of any activity, in and of itself, is not ordinarily considered of interest. However, when taken as a collective whole, a much different set of patterns emerges.
The behavior associated with activities (also called events) can be exposed using several different approaches. The goal is to take advantage of the similarities to understand the underlying behaviors associated with the activities being reviewed. Depending on the nature of the application, the behaviors will indicate different types of patterns that, once known, can then be identified and exploited according to the specific needs of the investigative agency. An analysis of activities can expose general trends or very specific patterns. Whether you are looking for temporal sequences or spatial dependencies, all information is contained within the activities themselves.
|
|
 |
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 >> |
 |
|
