Crime Detection via Crowdsourcing

Abstract. In this paper we propose a novel yet simple scheme for criminal detection. Rather than tracking the criminal that committed a particular crime, the police will rank the houses suspected to host criminals according to patterns on citizens’ tips. We show that this strategy will provably identify the desired houses under reasonable assumptions. This will aid detectives decide where to focus their efforts. We also give related problems of great interest to the community where the same ideas may be applied with similar results. We complement our theoretical findings with experiments that illustrate the effectiveness of this approach.