Finding Patterns with a Rotten Core: Data Mining for Crime Series with Core Sets

• In Event: Innovative Approaches for Assessing Crime at Place

Abstract

Detecting series' of crime is an important step in predictive policing, as knowledge of an ongoing pattern can be of paramount importance towards stopping it. Our goal is to assist crime analysts by providing automated tools for discovering crime series. Our approach relies on a key hypothesis that each crime series possesses a core of crimes that are similar to each other, which can be used to characterize the M.O. of the criminal. We find core sets of crime using an integer linear programming approach, and then construct the rest of the crime series by either merging core sets, or greedily grow from one core set, to form the full crime series. To judge whether a crime series is indeed a core, we consider both pattern-general similarity, which can be learned from past crime series, and pattern-specific similarity, which is specific to the M.O. of the series. We also learn a similarity graph on the set of crimes and build patterns based on this graph. Representing data in the form of a graph can potentially facilitate better and more nuanced understanding of serial patterns in the pattern general way and make pattern detection more computationally efficient.

Authors

• Tong Wang, Massachusetts Institute of Technology

• Cynthia Rudin, Massachusetts Institute of Technology

• Daniel Wagner, Cambridge Police Department

• Rich Sevieri, Cambridge Police Department