Suspects Investigation
In their efforts to identify potential suspects, crime investigators routinely draw on partial knowledge as the result of incomplete information and uncertain clues. Physical evidence gathered at a crime scene as well as accounts from victims and witnesse
- PDF / 154,055 Bytes
- 13 Pages / 439.36 x 666.15 pts Page_size
- 97 Downloads / 192 Views
Suspects Investigation
In their efforts to identify potential suspects, crime investigators routinely draw on partial knowledge as the result of incomplete information and uncertain clues. Physical evidence gathered at a crime scene as well as accounts from victims and witnesses may be incomplete and inconclusive. In cases with multiple offenders jointly committing a crime, where individual suspects have been identified, the aim of co-offending network analysis is to complement criminal profiling methods [3, 6] so as to identify additional suspects faster and more effectively, thus decreasing the cost and time of crime investigations. A common example is organized crime [4] as a form of criminal activity following a regular pattern, such as continuity or other spatiotemporal characteristics in contrast to irregular criminal behaviour associated with opportunistic crime. Law enforcement agencies often gain partial information about organized crime structures, for instance, from arrested suspects and convicted offenders who confess their affiliation with an organized crime group. However, uncovering the whole structure of criminal organizations under investigation poses a considerable challenge for law enforcement. Systematic approaches to co-offending network analysis can help tremendously in such cases. While many research studies in the literature use co-offending networks for crime suspect investigation, for instance [18, 19], to the best of our knowledge, none of these works define this problem formally by proposing an algorithmic solution. In this chapter, we introduce the problem of crime suspect recommendation as the goal to recommend the top-N additional potential suspects, given a partial set of crime suspects and a known co-offending network. To address this problem, we propose a random walk based method, called CRIMEWALKER, for link prediction and scenarios with a set of given suspects instead of a single source user. Section 5.1 explores related work, and Sect. 5.2 introduces a formal definition of the suspect investigation problem. Section 5.3 then presents CRIMEWALKER, while Sect. 5.4 describes the results of our experimental evaluation. Section 5.5 concludes this chapter. © Springer International Publishing Switzerland 2016 M.A. Tayebi, U. Glässer, Social Network Analysis in Predictive Policing, Lecture Notes in Social Networks, DOI 10.1007/978-3-319-41492-8_5
63
64
5 Suspects Investigation
5.1 Background Generally, crime analysis captures a broad spectrum of facets pertaining to different needs and using different analytical methods, namely: administrative analysis, strategic analysis, tactical analysis, criminal investigative analysis, and intelligence analysis. Administrative crime analysis aims at reducing or preventing crime by reporting local and regional statistics of crime rates to higher-ranking managers of law enforcement agencies. Strategic crime analysis primarily focuses on planning strategies for crime reduction and prevention. Tactical analysis tries to recognize repeating crime p
Data Loading...
