Feature Extraction from Internet Users' Comments for Early Detection of Depression

In this article, we extracted various features related to depression from the eRisk 2022 corpus. These features include, among others, the frequency of punctuation usage and the frequency of words with negative connotations. Such linguistic indicators can reveal individuals' emotional states and provide valuable clues for a better understanding of the textual manifestations of depression. An in-depth analysis of these elements aims to improve the automatic detection of psychological distress through language analysis.