NLNDE at SemEval-2023 Task 12: Adaptive Pretraining and Source Language Selection for Low-Resource Multilingual Sentiment Analysis

TL;DR

This paper presents a system for low-resource African language sentiment analysis using adaptive pretraining and source language selection, achieving top results in SemEval-2023 Task 12.

Contribution

It introduces language- and task-adaptive pretraining methods and source language selection strategies to improve sentiment analysis for low-resource languages.

Findings

Language-adaptive pretraining improves F1 scores by over 10 points.

Source language selection enhances transfer learning effectiveness.

The system wins 8 out of 15 tracks in the shared task.

Abstract

This paper describes our system developed for the SemEval-2023 Task 12 "Sentiment Analysis for Low-resource African Languages using Twitter Dataset". Sentiment analysis is one of the most widely studied applications in natural language processing. However, most prior work still focuses on a small number of high-resource languages. Building reliable sentiment analysis systems for low-resource languages remains challenging, due to the limited training data in this task. In this work, we propose to leverage language-adaptive and task-adaptive pretraining on African texts and study transfer learning with source language selection on top of an African language-centric pretrained language model. Our key findings are: (1) Adapting the pretrained model to the target language and task using a small yet relevant corpus improves performance remarkably by more than 10 F1 score points. (2) Selecting source languages with positive transfer gains during training can avoid harmful interference from dissimilar languages, leading to better results in multilingual and cross-lingual settings. In the shared task, our system wins 8 out of 15 tracks and, in particular, performs best in the multilingual evaluation.