BanglaSuicidalTextCorpus: A Corpus for Multi-Class Suicidal Risk Classification in Bengali Texts

A dataset of 5,100 labeled texts with n-gram embeddings gathered from Facebook, online blogs, YouTube comments etc. The data was categorized into three classes: no risk, low risk and high risk based on the suicidal sentiments on the text. Annotated by expert reviewers and cross-checked by three more, the dataset is valuable for analyzing suicidal sentiments, and developing advanced natural language processing applications in the domain of sentiment analysis, making it suitable for academic research and practical use in psychological medication.