Sentiment Classification on Suicide Notes Using Bi-LSTM Model

Sentiment classification has emerged as a key area of study in the current era, where sentiments and emotions play a significant role in all aspects of our daily lives, varying from personal to professional. This includes understanding people's opinions on various products as well as their feelings regarding a particular subject. Pertinent literature has shown that different methods and models have been used for sentiment classification on suicide notes. However, the state-of-the-art in this area was limited to using datasets with lower accuracies. Therefore, in this study, we developed a dataset using Quora, Reddit, several blogs, and various Infomedia sites for sentiment classification tasks on suicide notes. Moreover, we propose a Bi-LSTM model for the suicidal classification of a sentence into six categories such as ‘Love’, ‘Proud’, ‘Happy’, ‘Neutral’, ‘Sad’, and ‘Hate’. Our results show that the proposed model achieved a precision of 73%, a recall of 72%, and an F1-score of 72%.