A Social Recommendation Method Based on Double-Layer Weak Relation Network

In recent years, social recommendation become a popular research direction. Most social recommendation algorithms use solid relations for the recommendation, which causes a severe problem of accumulation of homogenized information in the recommendation list. Therefore, the paper proposes a social recommendation method based on a double-layer weak relation network. It can transmit heterogeneous information through a double-layer weak relation network. Firstly, the social recommendation method on weak relations reconstructs the knowledge graph and obtains the double-layer weak relations network. The first layered network uses SSLPA to establish a social trust network structure, the total weight of the network is quantified by the trust gate mechanism, and the migration confrontation technology determines the instance weight; The second layer is a weak relation social attribute network established by using the attribute information of nodes, and the social attribute network is constructed according to the bipartite graph technology. Secondly, the random walk model switches the double-layer weak relation network. After the random walk model reaches a stable distribution, the node heterogeneity recommendation list is obtained. Finally, building an experimental platform, and the experimental results show that the algorithm proposed in the paper has better heterogeneity propagation ability to ensure high accuracy.