PROJECT TITLE :

Reviewer Credibility and Sentiment Analysis Based User Profile Modelling for Online Product Recommendation

ABSTRACT:

Even for humans, deciphering user buying preferences, likes and dislikes is a difficult undertaking, making its automation a difficult assignment. To develop a robust recommendation approach, this research effort combines heuristic-driven user interest profiling with reviewer credibility analysis and fine-grained feature sentiment analysis. Candidate Feature Extraction, reviewer credibility analysis, user interest mining, candidate feature sentiment assignment, and recommendation module are the five modules of the proposed credibility, interest, and sentiment improved recommendation (CISER) model. The CISER model receives a review corpus as an input. The candidate feature extraction module extracts important characteristics based on context and sentiment confidence. Reviewer credibility analysis proposes an approach of associating knowledge, trust, and influence scores with reviewers to weigh their opinion according to their credibility to make our model robust to fraudulent and unworthy reviews and reviewers. The user interest mining module mines interest patterns using the aesthetics of review writing as heuristics. The candidate feature sentiment assignment module assigns a score to candidate features in a review based on the polarity of their FastText sentiment. Finally, for purchase recommendations, the recommendation module employs credibility weighted sentiment rating of user desired characteristics. For quantitative examination of numerous alternative items, the suggested recommendation approach incorporates not only numeric ratings, but also sentiment expressions connected with features, customer preference profiles, and reviewer credibility. CISER has a mean average precision (MAP@1) of 93% and a MAP@3 of 49%, which is higher than current state-of-the-art systems.