Crime analysis against women from online newspaper reports and an approach to apply it in dynamic environment
Priyanka Das, Asit Kumar Das · 2017
Crime against women in India has become an eminent topic of discussion in recent years and the issue has been brought to the foreground for concern due to the increasing trends in crimes performed against women. Most of the crimes get reported and a massive dataset is being generated every year. Analysing the crime reports can help the law enforcement section to take preventive measures for reducing the crime, but processing this voluminous data is strenuous and error-prone. So application of various text mining techniques can be of great help for visualising the crime trend. The present work proposes an effective methodology for analysing crime against women in India. The work involves extracting crime reports from online newspaper articles and the documents consisting crime reports of various states and union territories of India are made to undergo several preprocessing techniques. Each document is treated as bag-of-words and finally an exhaustive list of words have been prepared. Similarity has been measured among the words for selecting relevant features for crime trend analysis. Based on the selected crime features, community detection algorithm has been applied for partitioning the states based on crime against women in India. It is a graph based clustering approach and all the states of India have been considered as nodes of the graph. Each community is a group of states which are similar based on crime trends. The work also gives a new direction of application of the algorithm with certain limitations for incremental datasets where data are being generated regularly. The present methodology compares with some existing feature selection methods and clusters obtained in each method using same algorithm are evaluated which shows that the proposed method is more accurate than the existing methods.