Paper ID: 2401.12382

Longitudinal Sentiment Classification of Reddit Posts

Fabian Nwaoha, Ziyad Gaffar, Ho Joon Chun, Marina Sokolova

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.

Submitted: Jan 22, 2024