Web services recommendation based on Metapath-guided graph attention network

Existing Web services recommendation models suffer from matrix sparsity and cold-start problems due to the limited number of user-service interactions and too little contextual information, so this paper introduces knowledge graph as auxiliary information to Web service recommendation to alleviate matrix sparsity and cold-start problems. And we propose the Web services recommendation based on Metapath-guided Graph Attention Network Model (WSR-MGAT) to fully exploit the structural information of the knowledge graph to improve the recommendation accuracy. Specifically, WSR-MGAT uses graph embedding method to obtain the initial embedding of entities and relationships. In order to obtain more closely related neighbors, we propose to use a distance-aware path sampling method to extract meta-path instances with closer relationships. Previous knowledge graph-based Web service recommendations do not make full use of the rich interaction information, which may lead to limited performance. To address this problem, this paper uses meta-paths to guide nodes to recursively aggregate higher-order neighbor information and use an attention mechanism to distinguish the importance of neighbors. Meanwhile, the semantic information of nodes under different meta-paths is fused to obtain a more comprehensive embedding of nodes. To verify the performance of the model, we use the real data crawled from the ProgrammableWeb platform to conduct multiple groups of experiments. The experimental results show that the WSR-MGAT improves over the strongest baselines w.r.t. Prec@10 by 4.6%; Recall@10 by 3.8%; NDCG@10 by 4.2% and F1@10 by 4.9%.