MFSR: A novel multi-level fuzzy similarity measure for recommender systems

摘要

There is nowadays explosive growth and diversity of information due to the development of the internet. Thus, decision making in various fields has faced different challenges. Recommender systems by identifying the interests of users, data filtering and data management, offer personalized services to users. This is beneficial for marketing and user satisfaction. Recommender system has always faced challenges such as cold start, sparsity, scalability, accuracy, and quality. Collaborative Filtering (CF) as one of the most successful methods used in recommender systems is based on the similarity between users. We argue that similarity is a fuzzy notion and we get more realistic results in recommender systems by using fuzzy logic. Fuzzy logic deals better with uncertainty and is an effective method to identify ambiguities and uncertainty in measuring the similarity of items and users. In this paper, we present a new multi-level fuzzy similarity measure for recommender systems, called MFSR, which is based on popularity and significance. In order to improve the accuracy and quality of recommendations, we also propose a hierarchical structure for calculation of the similarity. To evaluate the contribution of this work, we use MAE, F1, recall, and precision. The MAE value based on the proposed similarity measure and the hierarchical structure is equal to 0.423 and outperforms the PIP and NHSM respectively by %4 and %13. Also, using the proposed similarity measure and the hierarchical structures, we obtain F1 value equal to 0.654, which outperforms the PIP and NHSM respectively by %17 and %20. We have also observed an improvement in recall and precision using the proposed approach. The results show that the proposed method (MFSR) performs better than similar methods in recent years such as PIP and NHSM.