On Importance of Code-Mixed Embeddings for Hate Speech Identification

TL;DR

This paper investigates the importance of code-mixed embeddings for hate speech detection, showing that HingBERT and Hing-FastText models trained on Hindi-English data outperform standard models on code-mixed hate speech datasets.

Contribution

It demonstrates the effectiveness of code-mixed embeddings and models trained on multilingual data for improving hate speech detection accuracy.

Findings

HingBERT outperforms BERT on hate speech detection in code-mixed data.

Hing-FastText surpasses standard FastText and vanilla BERT models.

Training on extensive Hindi-English data enhances model performance.

Abstract

Code-mixing is the practice of using two or more languages in a single sentence, which often occurs in multilingual communities such as India where people commonly speak multiple languages. Classic NLP tools, trained on monolingual data, face challenges when dealing with code-mixed data. Extracting meaningful information from sentences containing multiple languages becomes difficult, particularly in tasks like hate speech detection, due to linguistic variation, cultural nuances, and data sparsity. To address this, we aim to analyze the significance of code-mixed embeddings and evaluate the performance of BERT and HingBERT models (trained on a Hindi-English corpus) in hate speech detection. Our study demonstrates that HingBERT models, benefiting from training on the extensive Hindi-English dataset L3Cube-HingCorpus, outperform BERT models when tested on hate speech text datasets. We also found that code-mixed Hing-FastText performs better than standard English FastText and vanilla BERT models.