/* jquery */ /* jquery accordion style*/ /* jquery init */

7 July 2011

Technology for Law Enforcement

Inside my six page Technology for Law Enforcement feature article I uncover the increasing importance of technology in gathering evidence, solving crimes and preventing terrorist attacks.

As published in Micro Mart magazine issue 1165.

Some of the many technology-dependent areas I discuss are:
- social data trails
- video surveillance and car registration plate tracking
- facial tracking technology
- Eigenface facial recognition technology
- video surveillance
- operational data storage
- data mining software
- patterns and trends
- data visualisation
- geo-spatial information mapping

In the process I examine the benefits of various software products and services, such as Link Explorer and Indexer from Xanalys, Analyst's Notebook from i2 , ArcGIS from Esri, Google Maps and Microsoft Bing Maps.

Heres a couple of extracts from the article:

Retrieving information of interest from large data repositories is, rather aptly, referred to as data mining. Successful data mining is dependent on a set of carefully defined search criteria, accurately represented in well constructed data queries. Badly formed queries could result in investigations following inappropriate leads - very expensive in terms of time, manpower and financial resources.

Unfortunately, achieving this goal across the myriad of data repositories, often requires far more database knowledge and technical skills than is available to an investigation team. And yet it's only this team who know what queries are relevant to their lines of enquiry.

The solution? Ensure investigators can construct their own queries by investing in intuitive, graphically rich software data mining tools.

As you'd expect all the big database players possess comprehensive data mining tools. Oracle have developed a specialised Data Miner (ODM) product, supplied as an optional add-on. Microsoft incorporate data mining capability as part of their SQL Server Analysis Services.

Yet, by necessity these are general purpose products. Out-of-the-box they aren't as intuitive to use a many users would like. Tailoring them to an investigator's needs requires considerable customisation and configuration effort, which in turn demands high levels of technical knowledge and expertise.

However, there are alternatives. Alternatives that have been developed by software companies specialising in the law enforcement and investigative arena. Alternatives that aim to speak the same language as an investigator.

One such product is Link Explorer from UK-based Xanalys. Data mining queries are built using icons representing investigation-specific object classifications, such as people, locations, vehicles, locations, phone numbers, weapons, and time periods. Relationships are defined by graphically joining any number of these investigation-specific objects together. Finally, specific attributes constraints can be defined, for example only phone numbers that start '020'.

Using this combination of icons, graphical joins and custom constraints, it's a simple matter to construct the previously mentioned 'males between the ages of 20 and 40, who have an address in London's NW2 area and own a blue car'.

Another UK company with a long association with law enforcement organisations is i2, who also provide similar capabilities as part of their sizeable collection of analytical products.

As you might imagine in a busy investigation office, new items of information, connections, leads and human intuitions occur all the time. The beauty of these intuitive, interactive interfaces is they encourage the rapid generation of case-specific data mining queries.

When new pieces of intelligence come to light, such as a witness account of grey hair, or a change of detail such as a white car, the queries can be rapidly changed. They also enable speculative theories or 'hunches' to be quickly ascertained - for example possible associations with a particular person.

Though there's a case for all kinds of advanced software to help with information analysis, it's the 1.5kg of grey matter that sits between our ears, that's the most effective weapon in piecing together the pieces of a puzzle.

Textual documents, spreadsheets and paper-based reports have their uses, but aren't well suited to uncovering relationships, trends and highlighting coincidences. Instead, consider the traditional incident board covered with names, notes, photographs and interconnecting lines.

Such a board certainly makes a great visual prop for your TV murder mystery programme. But it also embodies a visual representation of easily digestible and mentally stimulating information. Just a single board can encapsulate an amazingly complex story, represented by an assemblage of interlinked people, events, locations, vehicles, items, times, and much more.

Imagine creating an electronic version of the incident board by displaying the output from all those investigation-specific data mining queries. A virtual board that could be any size, with zoom and scale control. Where the click of a mouse could reveal detailed data stored against the image or icon. Where the graphical layout and style could be instantly switched, to better highlight relationship strengths or sequences of time-based events.

There are a number of law enforcement focussed products that do just this. The previously mentioned Link Explorer from Xanalys and Analyst's Notebook from i2, render virtual incident boards in a variety of user-definable formats, layouts and styles. For anyone used to dealing with paper-based reporting systems of simple charts, the manner in which these tools bring the information to life and engage their grey matter can be a revelation.

As displays get ever bigger and cheaper, realistically sized virtual incident boards become more attainable. Many exhibit touch interfaces, encouraging interactive involvement from all team members. We're not quite able to achieve the gesture-based data manipulation scenario, as famously envisioned in the Minority Report film, but, with technologies like Microsoft Kinect it's actually possible to create a very good approximation of the concept.

Read more analysis posts.

No comments: