International Conference on Information and Knowledge Technology

فارسی

Home / شانزدهمین کنفرانس بین المللی فناوری اطلاعات و دانش

Evaluation of Machine Learning Models in Persian Text Classification for Depression Detection with Improved Feature Selection

Authors :

Rasool TayebMoghadam¹ Amirhossein Damia² Hadi Bahrampoor³ Maryam Zarghani⁴

1- وزارت آموزش و پرورش 2- دانشگاه صنعتی خواجه نصیرالدین طوسی 3- دانشگاه آزاد اسلامی واحد علوم و تحقیقات، یاسوج 4- دانشگاه صنعتی شاهرود

Keywords :

Persian Text Analysis،Depression Detection،Text Classification،Machine Learning

Abstract :

This study aims to identify and classify Persian texts related to depression using machine learning algorithms. The innovation of this research includes the collection and preparation of Persian data from various social networks such as Instagram, WhatsApp, Telegram, and Twitter. The dataset consists of 2000 Persian texts divided into two classes: "depression" and "non-depression." Features were extracted using the TF-IDF method, which was optimized through a custom approach. This optimization improved the accuracy of the machine learning models. Several algorithms, including SVM, Naive Bayes, KNN, and Random Forest, were evaluated. Results showed that the Naive Bayes algorithm, with an accuracy of 97.36%, outperformed other models. This research demonstrates that combining improved feature selection methods with various machine learning models can effectively identify depression in Persian texts and contribute to the development of high-accuracy text analysis systems.