Abstract:
Suicide is a major problem in latest civilization. Previously diagnosis and reduction of
suicide tries are crucial for saving people's lives. Suicidal thoughts can be detected using
clinical approaches involving social workers or specialists, as well as ML methods using
DL to identify online social material. This study is the first to introduce and describe the
approaches in these areas. In this regard, the issue of creating ways for detecting persons
who are predisposed to suicide conduct is becoming increasingly relevant. One of the
approaches of this research is to look for typological aspects of suicide-related speech
utilizing mathematical linguistics, computerized text processing, and machine learning. In
foreign science, texts written by persons who were motivated by suicide are researched
using automatic text processing methods, machine learning approaches, and models
designed to classify whether the content is connected to suicide or not. It should go without
saying that analyzing suicide notes together with other writings written by individuals who
have taken their own lives is essential to developing techniques for detecting individuals
who are at risk of suicide. There are two datasets such as suicidal and sentimental with total
2520 data and total 16 attributes. Passive-Aggressive, Random Forest, logistic regression,
SVM, MultinomialNB algorithms are using for sentiment dataset which find best PassiveAggressive (73%) accuracy and XGBClassifier (96%) accuracy for suicidal datasets.
