Applying particle swarm optimization algorithm-based collaborative filtering recommender system considering rating and review

With the rapid development of electronic commerce, the availability of a large amount of information on the products, as well as from other users, make the customers’ decision-making processes more time-consuming. The recommender system has emerged to assist the users choosing suitable products more easily, while companies can precision marketing more effectively. To solve the above-mentioned problems, this study adopted the particle swarm optimization algorithm (PSO) to determine the most suitable similarity of consumer ratings to avoid the problem of data distortion due to data sparsity. Moreover, bidirectional encoder representations from transformers (BERT) were applied to extract the characteristics of consumer feedbacks. Finally, the PSO was employed to determine the appropriate weight matrix and combine the characteristics of different data types. The combination of rating and review data could improve the recommendation performance. In addition, the proposed method was applied on six datasets of Amazon, and it outperformed several existing methods in terms of mean absolute error and mean squared error.