A Study of the Effects of Textual Features on Prediction of Terrorism Attacks in GTD Dataset.

Terrorist attacks are the biggest challenges to humanity around the world, which needs intensive efforts from researchers. Detecting the regularity of patterns and behaviors of terrorism is crucial to global counter-terrorism strategies. Machine learning techniques have shown significant effectiveness in the endeavor against terrorism. Nowadays, by using huge detailed terrorism data, researchers can develop tools that may contribute to dealing with terrorism. In this paper, we aim to create a framework for terrorism attacks predicting the use of global terrorism database (GTD). The research approach assumes that textual features may affect the enhancement of the classifier’s ability to predict the types of terrorist attacks. To prove this hypothesis; text features are extracted and represented using different text representation techniques such as Term Frequency-Inverse Document Frequency (TF-IDF), Bag of Words (Bow), and Word Embedding (W2vec). Extracted features are then combined with data set features, which are called (key features). Nine different classifiers are employed. The results show that the combination of textual features with key features improved the prediction accuracy significantly. [ABSTRACT FROM AUTHOR]