Eyewitness Information Management System Using Neuro-fuzzy Classification Schemes

Abstract

The use of computers in law enforcement has been generally recognized
and widely reported in the literature and recent popular press. Law
enforcement generates and deals with greater volume of data that takes a lot
of time to collate and work with. Previous successes with the use of software
tools in law enforcement have paved the way for new and innovative
applications of software techniques to aid crime data management and
investigation. The emerging area of homeland security is a particularly fertile
area of application for software-based investigative tools. Recent
developments in fuzzy systems and artificial neural networks have led to the
emergence of hybrid software systems that are effective for a variety of ill-
structured problems such as crime investigation. One solution to such
problems is based on a combination of expert rules, fuzzy logic reasoning,
prediction capabilities of artificial neural networks, and the data handling
power of computers. This paper presents the development of robust software
capable of utilizing and cross-referencing generic problems of witness
accounts, crime scene data, and investigators’ experiential reasoning. The
software, named EyeWitness Account (EWA), uses neuro-fuzzy classification
schemes for suspect identification based on eyewitness accounts.