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.
Abstract (translated)
我们报道了来自加拿大四大著名大学的学生在Reddit上撰写的帖子纵向情感分类的结果。我们专注于2020年至2023年的帖子文本。通过将情感阈值微调为[-0.075,0.075]的范围内，我们成功地构建了能够将帖子情感归类为积极和消极类别的分类器。值得注意的是，我们的情感分类结果在四个大学数据集上是一致的。
URL
https://arxiv.org/abs/2401.12382
