Cross-lingual Offensive Language Identification for Low Resource Languages: The Case of Marathi
This addresses the problem of offensive language detection for Marathi speakers and researchers, but it is incremental as it extends existing methods to a new language.
The paper tackled the lack of offensive language identification systems for low-resource languages by introducing MOLD, the first dataset for Marathi, and achieved results through machine learning experiments including zero-shot and transfer learning with cross-lingual transformers.
The widespread presence of offensive language on social media motivated the development of systems capable of recognizing such content automatically. Apart from a few notable exceptions, most research on automatic offensive language identification has dealt with English. To address this shortcoming, we introduce MOLD, the Marathi Offensive Language Dataset. MOLD is the first dataset of its kind compiled for Marathi, thus opening a new domain for research in low-resource Indo-Aryan languages. We present results from several machine learning experiments on this dataset, including zero-short and other transfer learning experiments on state-of-the-art cross-lingual transformers from existing data in Bengali, English, and Hindi.