HCMS at SemEval-2020 Task 9: A Neural Approach to Sentiment Analysis for Code-Mixed Texts

TL;DR

This paper presents a neural approach for sentiment analysis of code-mixed Hindi-English texts, demonstrating that simple convolution and attention mechanisms can achieve reasonable results despite resource limitations.

Contribution

The paper introduces a neural method using convolution and attention for sentiment classification in code-mixed texts, addressing resource scarcity issues.

Findings

Achieved an F1 score of 67.1% on the Sentimix Hindi-English task.

Showed that simple neural architectures can be effective for code-mixed sentiment analysis.

Highlighted the potential of transfer learning with limited resources.

Abstract

Problems involving code-mixed language are often plagued by a lack of resources and an absence of materials to perform sophisticated transfer learning with. In this paper we describe our submission to the Sentimix Hindi-English task involving sentiment classification of code-mixed texts, and with an F1 score of 67.1%, we demonstrate that simple convolution and attention may well produce reasonable results.