Hybrid Feature based Prediction of Suicide Related Activity on Twitter

Abstract

Suicide is a disturbing general medical issue and increasing fatal every year around the world. This work naturally removed casual inactive subjects from online webbased life twitter and communicating self-destructive ideations. Right off the bat emotionally assessed the idle points and afterward comprehensively contrasted them with chance variables proposed by space specialists. As long-range interpersonal communication destinations have gotten progressively normal, clients have embraced these locales to discuss strongly close to home points, among them their considerations about suicide. The tweets are significant for investigation since information shows up at a high recurrence and calculations that procedure them must do as such under extremely severe imperatives of capacity and time. Right now, we can separate Emoticons and Synonyms Feature and utilized ngram model which is a mix of Unigram, Bigram, and Trigram with half breed word reference for score computation. This model utilizing the casual points to anticipate the earnestness of the posts using machine learning algorithms. In this research, we also compare different approaches like SVM, NB, and RF.