Analyzing terror attacks using latent semantic indexing

Terrorism activities occur in many parts of the world. Such activities seemingly occur randomly in different locations, at different times, and are caused by different perpetrators. Thus, it is a challenging task to find patterns in activities related to terrorism. Clustering of perpetrators into similar groups can provide valuable information such as common characteristics among the various groups, the types of targets typically attacked, and weapons used in such attacks. In this research, we develop a method to classify terrorist groups based on their attack patterns by analyzing textual descriptions of such attacks using latent semantic indexing and clustering. The resulting information can be used for counter-terrorism globally, and that can help develop security measures that can be taken to protect potentially affected entities such as hospitals, schools, and government agencies as well as saving the lives of innocent people.