A Multilabel Multiclass Sentiment and Emotion Dataset from Bangladeshi E-Commerce Reviews

Published: 27 November 2023| Version 1 | DOI: 10.17632/rzjfg7t9kf.1
Mohammad Rifat Ahmmad Rashid, Kazi Ferdous Hasan, Md. Rakibul Hasan, Aritra Das, Mithila Sultana


This dataset provides a comprehensive view of the expanding online shopping landscape in Bangladesh, featuring customer reviews from popular e-commerce sites Daraz and Pickaboo. These reviews, a mix of Bengali and English text enriched with emojis, offer deep insights into customer sentiments towards widely favored products. The dataset undergoes thorough preprocessing and annotation for five emotions: Happiness, Sadness, Fear, Anger, and Love. These are further categorized into positive sentiments (Happy, Love) and negative sentiments (Sadness, Anger, Fear), facilitating a nuanced sentiment analysis. This dataset is a valuable asset for natural language processing research, offering a window into the diverse emotional responses of Bangladeshi online consumers and serving as a crucial tool for understanding sentiment in a multilingual, culturally rich market context



East-West University


Natural Language Processing, Sentiment Analysis