1 Matching and Predicting Crimes Dr. G.C. Oatley 1 , Prof. J. Zeleznikow 2 & Dr. B.W. Ewart 3 1. School of Computing and Technology, University of Sunderland, UK 2. School of Information Systems, Victoria University, Australia 3. Division of Psychology, University of Sunderland, UK Abstract Our central aim is the development of decision support systems based on appropriate technology for such purposes as profiling single and series of crimes or offenders, and matching and predicting crimes. This paper presents research in this area for the high-volume crime of Burglary Dwelling House, with examples taken from the authors’ own work a United Kingdom police force. Discussion and experimentation include exploratory techniques from spatial statistics and forensic psychology. The crime matching techniques used are case-based reasoning, logic programming and ontologies, and naïve Bayes augmented with spatio-temporal features. The crime prediction techniques are survival analysis and Bayesian networks. 1. Introduction The statutory requirement under the Crime and Disorder Act (1998) for United Kingdom police and local partnerships to undertake crime and disorder audits and produce strategies based on these audits, has provided a powerful stimulus to the mapping and analysis of crime data. [1] makes the point that the ‘recent shift within British policing towards a more decentralised, proactive style has shifted the analytical focus onto analysts and intelligence officers at the police divisional level who are now expected to be the hub of the local intelligence gathering effort. For high volume crime, this has left an analytical void.’ The authors work is in this area is with the high volume crime of Burglary from Dwelling Houses (BDH) though collaboration with West Midlands Police (WMP). Software [2] was developed to interrogate the database of recorded crimes in order to explore the temporal and spatial characteristics of BDH across the entire operational command unit. The objectives were