Police Report Linking Algorithm Based on Named Entity Recognition

Abstract

Police reports include a complaint made by the victim of a crime. This complaint is a story that describes the facts from the victim’s point of view. Most of the time, crimes are perpetrated by unknown authors. This causes cases to be shelved until new evidence arrives. The use of natural language processing allows us to take advantage of non-structured text in victim’s com plaints by generating links that could lead to the reopening of an archived investigation. Through the use of NER, it is possible to extract entities of interest from a report of a complaint that arrives and link it with other reports of exist ing complaints, allowing the generation of a maps of links in order to detect similarities between cases. This idea will increase the possibility that cases in archived status being opened.