An Effective Method to Detect Serial Crimes
In this paper, we propose models to predict the serial murders anchor point and next crime location. First, we attempt two ways, Circle Hypothesis and The Least-effort Principle, to search the anchor point. Then we use the Fuzzy C-Means Method to classify all the historical crime locations into several regions. According to these regions, we generate a geographical profile which is the most possible location of the offenders residence using the Normal Distribution Model. All the methods come to a same conclusion. Whats more, according to the close relationship between criminal scenes and offenders residence, we establish a model to predict the next possible crime scene mainly decided by previous crime sites and time interval. In the model, we draw a high probability area depending on previous crime sites. At the same time, we take account into some other factors to optimize this model in real situations, such as the population density, the local geography and demographic backcloth, time interval, the success rate of committing in this region and so on. In the case of Peter Sutcliffe, We get the conclusion that the offenders residence is in Bradford, which is fit to the actual situation. At the same time we predict the last 16 crime sites, among which 10 are accurate. The accuracy rate is 62.5％.
Geographical Profile The Least-effort Principle FCM Normal Distribution
Ze Yang Jing Wang Lu Liu Jingyuan Huang
College of Atmosphere Physics, Nanjing University of Information Science & Technology College of Atmosphere Science, Nanjing University of Information Science & Technology College of Computer and Software, Nanjing University of Information Science & Technology
国际会议
The Third International Conference on Modelling and Simulation(第三届国际建模、计算、仿真、优化及其应用学术会议 ICMS 2010)
无锡
英文
203-206
2010-06-04（万方平台首次上网日期，不代表论文的发表时间）