Abstract
In this paper, we present our system for the SemEval 2020 task on code-mixed sentiment analysis. Our system makes use of large transformer based multilingual embeddings like mBERT. Recent work has shown that these models posses the ability to solve code-mixed tasks in addition to their originally demonstrated cross-lingual abilities. We evaluate the stock versions of these models for the sentiment analysis task and also show that their performance can be improved by using unlabelled code-mixed data. Our submission (username Genius1237) achieved the second rank on the English-Hindi subtask with an F1 score of 0.726.- Anthology ID:
- 2020.semeval-1.122
- Volume:
- Proceedings of the Fourteenth Workshop on Semantic Evaluation
- Month:
- December
- Year:
- 2020
- Address:
- Barcelona (online)
- Venue:
- SemEval
- SIG:
- SIGLEX
- Publisher:
- International Committee for Computational Linguistics
- Note:
- Pages:
- 951–956
- Language:
- URL:
- https://aclanthology.org/2020.semeval-1.122
- DOI:
- 10.18653/v1/2020.semeval-1.122
- Cite (ACL):
- Anirudh Srinivasan. 2020. MSR India at SemEval-2020 Task 9: Multilingual Models Can Do Code-Mixing Too. In Proceedings of the Fourteenth Workshop on Semantic Evaluation, pages 951–956, Barcelona (online). International Committee for Computational Linguistics.
- Cite (Informal):
- MSR India at SemEval-2020 Task 9: Multilingual Models Can Do Code-Mixing Too (Srinivasan, SemEval 2020)
- PDF:
- https://preview.aclanthology.org/starsem-semeval-split/2020.semeval-1.122.pdf