An Aspect-Based Sentiment Analysis Dataset of Skincare Product Reviews from Bangladeshi E-Commerce Platforms
Description
This dataset presents an aspect-level sentiment annotated corpus of skincare product reviews collected from major Bangladeshi e-commerce platforms (Shajgoj, Ogerio, and Ohsogo). It comprises 2,003 aspect-level annotated instances, developed to support Aspect-Based Sentiment Analysis (ABSA) in the skincare domain. The dataset captures fine-grained consumer opinions by associating specific product aspects with corresponding sentiment polarity. The data is structured into the following fields: Product Name: Name of the skincare product Product Category: Category of the product (e.g., Sunscreen) Rating: User-provided rating on a 1–5 scale Review Date: Date when the review was posted Review Text: Customer-written review content Aspect-Sentiment Annotation: Aspect category paired with sentiment polarity in JSON format The dataset covers seven aspect categories—Suitability, Effectiveness, Quality, Texture, Sensitivity, Satisfaction, and Others—with sentiment labels classified as Positive or Negative. Reviews containing multiple aspects are split into separate aspect-level entries. The corpus was curated from real user reviews and manually annotated to ensure consistency and labeling accuracy. Applications: Aspect-Based Sentiment Analysis Opinion Mining Text Classification Consumer Behavior Analysis Explainable AI in E-commerce
Files
Steps to reproduce
Download the dataset from Mendeley Data. Load the dataset using spreadsheet software or data analysis tools such as Python or R. Extract the review text and aspect–sentiment annotations. Apply standard text preprocessing and use the data for sentiment analysis or aspect-based sentiment analysis tasks.
Institutions
- University of Brahmanbaria
- Daffodil International University
- Bangladesh University of Business and Technology
- Jahangirnagar University