Multi-Label Sentiment Analysis on 100 Languages with Dynamic Weighting for Label Imbalance

TL;DR

This paper presents a novel multi-label sentiment analysis framework for 100 languages that uses dynamic class weighting and optimized thresholds, achieving state-of-the-art results across multiple languages and metrics.

Contribution

It introduces a dynamic weighting method and an efficient threshold selection technique for multi-label sentiment analysis across many languages, improving over static methods and previous benchmarks.

Findings

Achieves state-of-the-art performance in 7 out of 9 metrics across 3 languages.

Effectively handles class imbalance with dynamic weighting.

Supports sentiment analysis in 100 languages with a single model.

Abstract

We investigate cross-lingual sentiment analysis, which has attracted significant attention due to its applications in various areas including market research, politics and social sciences. In particular, we introduce a sentiment analysis framework in multi-label setting as it obeys Plutchik wheel of emotions. We introduce a novel dynamic weighting method that balances the contribution from each class during training, unlike previous static weighting methods that assign non-changing weights based on their class frequency. Moreover, we adapt the focal loss that favors harder instances from single-label object recognition literature to our multi-label setting. Furthermore, we derive a method to choose optimal class-specific thresholds that maximize the macro-f1 score in linear time complexity. Through an extensive set of experiments, we show that our method obtains the state-of-the-art performance in 7 of 9 metrics in 3 different languages using a single model compared to the common baselines and the best-performing methods in the SemEval competition. We publicly share our code for our model, which can perform sentiment analysis in 100 languages, to facilitate further research.