Longitudinal Sentiment Classification of Reddit Posts

Abstract

We report results of a longitudinal sentiment classification of Reddit posts written by students of four major Canadian universities. We work with the texts of the posts, concentrating on the years 2020-2023. By finely tuning a sentiment threshold to a range of [-0.075,0.075], we successfully built classifiers proficient in categorizing post sentiments into positive and negative categories. Noticeably, our sentiment classification results are consistent across the four university data sets.

arXiv comment11 pages, 10 figures, 4 tables