Aspect-based sentiment analysis based on users' comments in an online marketplace

Articles in Press

Document Type : Computer Article

Authors

1 Department of Computer Engineering, Yazd University, Yazd, Iran

2 Department of Computer Engineering, Yazd University

Abstract

In recent years, with the expansion of online shopping and the importance of user feedback in improving the quality of products and services, sentiment analysis of user reviews has become one of the most important tools and opportunities for online businesses and services. In this research, various approaches based on Convolutional Neural Networks (CNN), BERT language model, and FastText language model have been examined for sentiment analysis of user reviews about mobile phones in aspects such as camera, battery, and price in an online marketplace. In this regard, the data were labeled based on aspects and categorized into three sentiments: positive, negative, and neutral. Additionally, to improve the performance of the proposed approaches and address data imbalance, data augmentation methods were utilized and their impact was analyzed. Finally, using the proposed models, very high accuracy in detecting positive, negative, and neutral reviews in each aspect was achieved. According to the results, the proposed CNN-based approach performed better than the other two proposed methods on the given data.

Keywords

Main Subjects

Journal of Modeling in Engineering

Articles in Press, Accepted Manuscript
Available Online from 14 September 2025

Files

History

  • Receive Date: 28 April 2024
  • Revise Date: 23 March 2025
  • Accept Date: 01 June 2025

Share

How to cite

Statistics