An Opinion Mining-Based Hybrid Collaborative Filtering Recommendation Model for Consumer Decision

Abstract

A variety of information sources can be utilized by a recommendation system to suggest items that align with the diverse interests of its users. Typically, such systems employ the collaborative filtering (CF) method, which involves combining a user's preference data with that of others to anticipate additional items that may pique their interest. This research proposes a novel weighted recommendation system based on CF to facilitate better decision-making among consumers. The study presents two equations, one for calculating the weight of a product and its review, and another for determining the similarity between reviews provided by different consumers. The methodology utilizes a combination of Random Forest (RF), Support Vector Machine (SVM), and Logistic Regression (LR) ensemble classifiers to improve model performance. The proposed model is trained and evaluated using an open-source dataset available on Kaggle's website. The numerical analysis indicates that the proposed model outperforms other conventional methods in terms of accuracy (0.821), precision (0.802), recall (0.821), F-measure (0.833), and error rate (0.100), among other metrics.