Intelligent Data Acquisition Framework of Terrorist Attacks Using Natural Language Processing

Abstract

Eradicating terrorism using state of the art technologies has proved to be an active area of research with the increase in the incidents involving terrorism. It has been observed that the terrorism incident data available has played an important role in devising the techniques to help curb the terrorism. Therefore, the importance of acquiring the terrorist incident data is very crucial part. The gathered data gives researcher the ability to identify and link the patterns used by terrorist through the analysis of gathered data. This is done using different Machine learning algorithms and therefore helps in the prevention of future terrorist activities. However, due to the unavailability of an automated system the terrorism incident data collector has to perform a very lengthy task of going through different news articles to gather data and also has to verify the accuracy of the data. This makes it very hard to update the database in time. In this research, we have taken the advantage of Natural Language Processing and specifically Named Entity Recognition to obtain entities from news articles containing reports of terrorist incidents. We have specifically trained the opensource library of Python named as Spacy to perform this task, which used Convolutional Neural Network Classifier to train identify entities in a text. Presently, no approach is powerful enough to create a fully automated database creation process with 100% accuracy, but our approach significantly reduces the overhead of the terrorist incidents data collection process. This will ultimately lead to a more convenient and fast way of terrorism incidents collection of data.