The Use of Hypergraph for Collaborative Filtering Recommendation Method

Currently, the number of users using online platforms to purchase various products and services is increasing. Recommender systems allow users to find easier products and services based on their preferences and tastes. This paper presents a collaborative recommendation system using a hypergraph network structure. These systems create a profile of the target user using their similarity to some other ones. It is the reason, the collaborative recommendation method is susceptible to the measure of similarity which is utilized to express the dependence among users and products (products). To better express the interlinkages among users and products in the recommender system, we suggest a recommendation algorithm for collaborative filtering which is based on the similarity measure in hypergraphs and further using the clustering method. Using a hypergraph to model user interactions with products makes it possible to analyze user groupings and create recommendations for individuals within each community. This paper introduces a hypergraph model for capturing complex relationships, describes the necessary algorithms, and ability to produce appropriate recommendations based on experimental data. Experiments were performed on two standard datasets (MovieLens100k and CiaoDVD). The results showed that the suggested method is applicable and gives comparable results to other collaborative recommendation algorithms.