Quantitative Analysis of Terrorist Attack Data Based on Weighted Clustering
Abstract— Based on the terrorist attacks of 1998-2017 in GTD, a global terrorism database, this paper makes a deep analysis of the data related to terrorist attacks. By preprocessing and quantifying the data, it mainly uses the theoretical model of factor analysis. Classifying terrorist attacks according to many factors of harmfulness, then analyzing the data between multiple terrorist attacks, using the informat ion of known perpetrators, according to the hazardous screening of suspects, using the characteristics of terrorist attacks, The correlation analysis model is used to determine the unknown perpetrators in terrorist attacks, and the similarity measurement model is used to determine the suspect degree of the perpetrators. The size of The details are as follows: Through pre-analysis of the data, we can f ind out the main determinants of the harmfulness of terrorist attacks, including casualties, economic losses, timing, location, and so on. Through the analysis of the degree of influence of the factors, A model of clustering weight hazard classification based on factor analysis theory is established. Factor analysis weight method is used to solve the characteristic weight vector in data analysis and processing. Firstly, clustering is used to classify the terrorist attack events with similar harmfulness assessment, then the overall harmfulness of in-class events is quantitatively evaluated, and the harm of terrorist attacks is divided from low to high into one to five grades. The rating of future terrorist attacks is based on the distance between the event characteristics and the endoplasmic center. The new event is classified to the nearest centroid class, and the model automatically updates the centroid. According to the hierarchical model, the ten most harmful terrorist attacks in the past twenty years can be screened. In a sense, it can be very close to the comprehensive subjective division of many people. The theory of cosine similarity is introduced on the basis of the previous experiment, and the correlation model based on mult idimensional vector cosine similarity is established. First of all, by using the informat ion of the known perpetrators, we exclude the data from which the perpetrators have been arrested, screen out terrorist organizations or individuals according to the magnitude of harmfulness, fu rther analyze the characteristics of terrorist organizations that cause terrorist attacks, and generate feature vectors. At the same time, it analyzes the characteristics of the attack events that no organization or individual claims to be responsible for, and generates the feature vectors. Finally, the multi-dimensional vector cosine similarity algorithm is used to solve the suspect degree of the event suspect organization. Screen out the most likely perpetrators of each incident.
Index Terms— Quantitative Analysis, Factor Analysis weight method, clustering, CoSine similarity
Lei Juchao, Wang Yaming
School of Computer Science and Engineering Xi'an Technological University, CHINA
Cite: ei Juchao, Wang Yaming, "Quantitative Analysis of Terrorist Attack Data Based on Weighted Clustering," Proceedings of 2019 the 9th International Workshop on Computer Science and Engineering WCSE_2019_SPRING, pp. 63-67, Yangon, Myanmar, February 27-March 1, 2019.