Towards Code-switched Classification Exploiting Constituent Language Resources
This work addresses the challenge of limited data for code-switched classification, specifically for tasks like sarcasm and hate speech detection in English-Hindi, but it is incremental as it applies an existing conversion method to new tasks.
The paper tackled the problem of analyzing code-switched data, which is limited in availability, by converting it into its constituent high-resource languages to exploit monolingual and cross-lingual resources. The result showed a 22% increase in F1-score for sarcasm detection and a 42.5% increase for hate speech detection compared to state-of-the-art methods in English-Hindi code-switched settings.
Code-switching is a commonly observed communicative phenomenon denoting a shift from one language to another within the same speech exchange. The analysis of code-switched data often becomes an assiduous task, owing to the limited availability of data. We propose converting code-switched data into its constituent high resource languages for exploiting both monolingual and cross-lingual settings in this work. This conversion allows us to utilize the higher resource availability for its constituent languages for multiple downstream tasks. We perform experiments for two downstream tasks, sarcasm detection and hate speech detection, in the English-Hindi code-switched setting. These experiments show an increase in 22% and 42.5% in F1-score for sarcasm detection and hate speech detection, respectively, compared to the state-of-the-art.