Mining Criminal Databases to Finding Investigation Clues—By Example of Stolen Automobiles Database

摘要

While businesses have been extensively using data mining to pursue everlasting prosperity, we seldom consider this technique in public affairs. The government holds a large quantity of data that are records of official operations or private information of the people. These data can be used for increasing benefits of the people or enhancing the efficiency of governmental operations. In this paper we will apply this technique to the data of stolen automobiles to explore the unknown knowledge hidden in the data and provide this knowledge to transportation, insurance as well as law enforcement for decision supports. The data we use are abstracted from 378 thousand records of stolen automobiles in the past eleven years in Taiwan. After constructing a data warehouse, we apply the technique of classification, association rule, prediction, data generalization and summarization-based characterization to discover new knowledge. Our results include the understanding of automotive theft, possibility of finding stolen automobiles, intrigue in theft claims, etc. The knowledge we acquired is useful in decision support, showing the applicability of data mining in public affairs. The experience we gathered in this study would help the use of this technique in other public sectors. Along with the research results, we suggest the law enforcement to consider data mining as a new means to investigate criminal cases, to set up a team of criminal data analysis, to launch a new program to crack down automotive thefts, and to improve the quality of criminal data management.