In order to investigate the prospect of employing machine translation to create a parallel multilingual hate speech dataset, the researchers have designed a scientific pipeline. By evaluating the quality of the translations, determining the toxicity levels of the source and target texts, and computing correlations between the newly acquired scores, the research analyzes how/whether this is possible. To obtain further grammatical and semantic understanding, the investigators lastly conduct a qualitative study. By analyzing the difficulties of the work, they want to investigate methods of filtering hate speech texts in order to parallelize sentences in many languages.https://aclanthology.org/2024.lrec-main.1376Share this:FacebookXLike this:Like Loading... Post navigation SAFE-MEME: Structured Reasoning Framework for Robust Hate Speech Detection in Memes (arXiv)