Text classification for detection of potential threat

Fatima Chiroma
PhD student

The social network has become a vital medium for communication. People have the freedom to share their opinions using different online social network platforms. This has however led to some growing concerns regarding its negative impact on human safety, for example cyberbullying and the spread of suicidal ideation.

Therefore this work focuses on the application of machine classifiers: Support Vector Machine (SVM), Naive Bayes (NB), Decision Tree (DT) and Random Forest (RF), on suicide-related tweets, and their accuracy in identifying and distinguishing suicidal communications. The results of the experiment show that SVM has the best score for Precision, while NB has the best Recall score, F-measure and Accuracy. Overall, NB has slightly outperformed the other classifiers in three of the four standard classification measures: Recall, F-measure and Accuracy.

This entry was posted in SC2018 and tagged , , . Bookmark the permalink.