The article explores the issue of temporal bias in abusive language detection across various languages. This bias can lead to a decrease in the accuracy of detection models, potentially allowing abusive language to go undetected or be falsely detected. The study evaluates the performance of models on abusive data sets from different time periods, demonstrating that temporal bias is a significant challenge. The authors also provide a linguistic analysis of these datasets from a diachronic perspective, aiming to explore the reasons for language evolution and performance decline. The study provides crucial insights into language evolution and ways to mitigate temporal bias.
Publication date: 26 Sep 2023
Project Page: N/A
Paper: https://arxiv.org/pdf/2309.14146
