A Lexicon-Based Approach for Hate Speech Detection

This paper addresses several of the key issues facing creation of a classifier for hate-speech on forums, blogs, or other areas of web discourse. The researchers in this article define the content of hate speech into three subcategories: race, nationality, and religion, and analyze responses for their subject and opinion. As this research is preliminary, it deals with the capacity for a model to detect subject and to determine the expressed sentiment on a scale. The experiments conducted by the researchers find that the inclusion of semantic, hate and theme-based features produces the best results in lexicon sentiment analysis, while the usage of subjective sentiments also improved performance of the model.

This article will be most useful to researchers, practitioners and professionals that seek to create a similar sentiment analysis tool or to improve upon an existing model.