A knowledge discovery approach to supporting crime prevention

Abstract

Recently, the increasing volume crimes have been one of the most serious issues in Taiwan. The Ministry of Internal Affairs has been working out a project to examine the public security index with red, yellow, purple, green, blue five lights in order to strengthen the public security. For analyzing and predicting huge linguistic data which evolve with time, we propose a novel fuzzy self-organization map network to uncover crime trend and use association rule to discover the hidden causal effects between two different criminal time series data. The fuzzy self-organization model integrates the features of dealing with clustering and linguistic data of SOM and fuzzy logic, respectively. We analyze the clustering results on distinguishing different trends of each criminal category and find rules relating patterns in a time series to other patterns, for the purpose of specifying the instructions of police human resources planning. The resulting findings can facilitate the development of a useful decision-support tool for helping decision makers determine appropriate law enforcement strategies.