| Suicidal Ideation Detection on Social Media Using A Hybrid Feature Selection Method |
| Paper ID : 1056-ICEEM2023 (R1) |
| Authors: |
|
Abdallah Basyouni *1, Hatem Abdelkader2, Asmaa Ali3 1Information Systems Department, Faculty of Computers and Information, Menoufia University, Shebin Elkom, Egypt 2Information Systems Department, Faculty of Computers and Information, Menofia University 3Information Systems Department, faculty of computer and information, Menoufia University |
| Abstract: |
| Suicide is a major issue in modern communities all around the world. Suicide can be caused by several risk factors such as depression, anxiety, hopelessness, and social isolation. The early discovery of these risk factors can decrease a lot of suicide attempts. In later years, suicidal ideation detection via online social media has risen, and become a hot research topic with significant challenges in both fields of natural language processing (NLP) and psychology. This paper proposes a framework to detect suicidal ideation on social media. The proposed framework consists of four phases, preprocessing, feature extraction, feature selection, and classification. In the feature selection phase, the paper introduces a hybrid feature selection method combining principal component analysis (PCA) with genetic algorithm (GA) to select the best and most important features. Moreover, an expressive feature set was extracted based on context-related and linguistic features. The proposed hybrid feature selection method performs better and achieved an accuracy of 99.05%. This paper shows that a more accurate classification procedure may be obtained by using expressive feature sets and choosing relevant and informative features |
| Keywords: |
| Suicide Ideation, Feature Selection, Natural Language Processing, Principal Component Analysis, Genetic Algorithm |
| Status : Paper Accepted |
