Research publication · FIRE 2021 Working Notes · CEUR-WS
Deep Learning Based Sentiment Analysis for Malayalam, Tamil and Kannada Languages
Proceedings publication:
My credited role: First-listed author
The research problem
Code-mixed text combines languages within the same comment. Our team studied sentiment classification for Malayalam-English, Tamil-English and Kannada-English in the Dravidian-CodeMix shared task at FIRE 2021.
Approach and attribution
The paper compares dense neural networks, bidirectional LSTMs and a CNN–LSTM model, and discusses preprocessing and class imbalance.
I am credited as the first-listed author, Pavan Kumar P.H.V., alongside Premjith B, Sanjanasri J.P and Soman K.P. We submitted this work as team Amrita_CEN_NLP.
Paper and code
- Read the published paper (PDF).
- Explore the team’s public research code, also linked from the paper.
- View the proceedings and publication date.
FIRE 2021 is the event year. CEUR-WS published this proceedings volume on 1 July 2022, so the portfolio indexes the paper under 2022.