ECMAG---An-Ensemble-Framework-for-Sentiment-Analysis-in-Code-Mixed-Data

This repository contains the code for the paper titled "ECMAG - Ensemble of CNN and Multi-Head Attention with Bi-GRU for Sentiment Analysis in Code-Mixed Data" which has been accepted at the FIRE 2021: Forum for Information Retrieval Evaluation, India.

The work proposes an ensemble framework Ensemble of Convolutional Neural Network and Multi-Head Attention with Bidirectional GRU (ECMAG)to map the code-mixed user comments to their corresponding sentiments. The perfomance of the framework was tested on the Tamil-English code mixed dataset. The work is part of the Dravidian CodeMix FIRE 2021 – Sentiment Analysis for Dravidian Languages in Code-Mixed Text task.

Requirements can be installed by entering the following command in the Terminal:

pip install -r requirements.txt

Publication

You can find our paper [here] (yet to be published).

Authors

Prasannakumaran D, Sideshwar J B, Thenmozhi D