Knowledge modelling in law
National Criminal Justice Modelling Workshop: 2003
Presenter: Professor John Zeleznikow, Victoria University
Overview of presentation:
We are conducting current research under the framework of the CAST project to develop intelligent decision support systems that use both expert knowledge and automated data analysis to guide analysts and investigators in the identification of complex criminal networks.
In this paper we first investigate User requirements for Decision Support for Criminal Investigation and Principles for constructing Criminal Investigation Decision Support Systems.
We then introduce the FLINTS methodology and software system. FLINTS was developed to support the detection of high volume crimes within the West Midlands, through a judicious choice of queries to evidential databases of DNA, fingerprints, footwear and tool-marks. Analysis of the data reveals patterns, associations and links which would not have been detected had each evidence type been managed in separate systems.
FLINTS is a new approach to knowledge management in that it releases the inherent power in large data collections used by law enforcement. Through a judicious choice of questions, knowledge about scenarios, links, stories and connections between many types of data and many types of events as well as many people and many locations can be inferred. Results are visualized to aid analysts understand the chains of links and then contemplate new searches for new links. FLINTS is therefore currently a decision support system for analysts and investigators that helps them identify relevant information amongst a mass of data. The strength of the system is the identification of what should be ‘obvious’ links between people and crimes but are hidden in mixed masses of data.
In this paper we describe how we have extended the FLINTS methodology with enhanced data visualisation tools to help crime investigators focus upon relevant data. Many crime detection decision support systems use sophisticated techniques (such as Bayesian networks or data mining) to make decisions about the perpetrators of crime. The purpose of the CAST paradigm is to help improve the performance of crime investigators and not replace their critical faculties. The methodology focuses upon identifying links between criminal acts, criminal actors and their locations. The resulting software has been tested in various areas of financial fraud, including car insurance fraud, VAT abuse and investment fraud.
We conclude by examining how our techniques have been used in civil law domains—particularly property distribution in Australian Family Law.
| Release | PDF Size | |
|---|---|---|
| Knowledge modelling in law | (80 kB) |
If you need help with any of the file types above, please see our help page.
More Information
- For further information about this page please use the Request a Statistic form or phone (07) 3224 5284.
Last reviewed: Dec 19, 2006, Last modified: Dec 13, 2006
Other languages