This paper presents a comparative analysis of noise reduction methods in sentiment analysis on noisy Bengali texts. It introduces a manually annotated dataset (NC-SentNoB) to identify ten different types of noise in Bengali texts. The paper addresses the identification of noise type as a multi-label classification task and presents baseline noise reduction methods. The study concludes that the current noise reduction methods are not satisfactory, thus highlighting the need for more suitable techniques in the future.
Publication date: 26 Jan 2024
Project Page: https://github.com/ktoufiquee/A-Comparative-Analysis-of-Noise-Reduction-Methods-in-Sentiment-Analysis-on-Noisy-Bengali-Texts
Paper: https://arxiv.org/pdf/2401.14360
