Welcome to the InfoSci Platform
Reference Hub1
A Hybrid Approach of Prediction Using Rating and Review Data

A Hybrid Approach of Prediction Using Rating and Review Data

Aseem Srivastava, Rafeeq Ahmed, Pradeep Kumar Singh, Mohammed Shuaib, Tanweer Alam
Copyright: © 2022 |Volume: 12 |Issue: 2 |Pages: 13
ISSN: 2155-6377|EISSN: 2155-6385|EISBN13: 9781683182092|DOI: 10.4018/IJIRR.299942
Cite Article Cite Article

MLA

Srivastava, Aseem, et al. "A Hybrid Approach of Prediction Using Rating and Review Data." IJIRR vol.12, no.2 2022: pp.1-13. http://doi.org/10.4018/IJIRR.299942

APA

Srivastava, A., Ahmed, R., Singh, P. K., Shuaib, M., & Alam, T. (2022). A Hybrid Approach of Prediction Using Rating and Review Data. International Journal of Information Retrieval Research (IJIRR), 12(2), 1-13. http://doi.org/10.4018/IJIRR.299942

Chicago

Srivastava, Aseem, et al. "A Hybrid Approach of Prediction Using Rating and Review Data," International Journal of Information Retrieval Research (IJIRR) 12, no.2: 1-13. http://doi.org/10.4018/IJIRR.299942

Export Reference

Mendeley
Favorite Full-Issue Download

Abstract

A collaborative filtering technique has proven to be the preferable approach for personalized recommendations. Traditionally, collaborative filtering recommends target items to those users who have similar tastes. The performance of collaborative filtering degrades significantly when a considerable number of users do not provide ratings on recommended products. In such a scenario, the dataset utilized in recommendation becomes highly sparse, and ratings become very few or none co-rated. To mitigate the problem, as mentioned earlier, and to improve the performance of collaborative filtering, we propose an approach that adopts users' textual reviews and ratings both in the rating prediction. The dataset used is Amazon fine Food Reviews containing rating and text review with 568454 reviews from October 1999 to October 2012. The proposed model is tested on the collected dataset. The experimental results provide the proper evidence that the proposed model outperforms other traditional algorithms of collaborative filtering techniques.