E-commerce Product Recommendations Based on User Review and Collaborative Filtering

Abstract

The fast growth in e-trade industry has resulted to the introduction on recommendation systems that makes useful suggestions and assists clients in accessing the lengthy tail merchandise. Recommendation structures employ the use of client feedbacks to make product suggestions. The conventional advice systems typically rely upon product scores given by means of clients. However, with the development in records acquisition, e-commerce web sites are capable of retrieving useful feedbacks. This paper tries to enhance the performance of recommendation systems based on client review. This method employs collaborative filtering, sentiment analysis, K-means algorithm to make product recommendations. Flipkart product dataset consisting of diverse product ratings and reviews are utilized to evaluate the performance of proposed method. The result of the experiment proves that the proposed method outperforms the traditional rating-based recommendation machine.