BAKSA at SemEval-2020 Task 9: Bolstering CNN with Self-Attention for Sentiment Analysis of Code Mixed Text

TL;DR

This paper introduces an ensemble model combining CNN and self-attention LSTM for improved sentiment analysis of code-mixed tweets, achieving competitive F1 scores on Hinglish and Spanglish datasets.

Contribution

It presents a novel ensemble architecture leveraging CNN and self-attention LSTM specifically for code-mixed sentiment analysis, enhancing classification accuracy.

Findings

Achieved F1 scores of 0.707 on Hinglish

Achieved F1 scores of 0.725 on Spanglish

Ranked 5th and 13th in respective tasks

Abstract

Sentiment Analysis of code-mixed text has diversified applications in opinion mining ranging from tagging user reviews to identifying social or political sentiments of a sub-population. In this paper, we present an ensemble architecture of convolutional neural net (CNN) and self-attention based LSTM for sentiment analysis of code-mixed tweets. While the CNN component helps in the classification of positive and negative tweets, the self-attention based LSTM, helps in the classification of neutral tweets, because of its ability to identify correct sentiment among multiple sentiment bearing units. We achieved F1 scores of 0.707 (ranked 5th) and 0.725 (ranked 13th) on Hindi-English (Hinglish) and Spanish-English (Spanglish) datasets, respectively. The submissions for Hinglish and Spanglish tasks were made under the usernames ayushk and harsh_6 respectively.