Deep learning model for recommendation system using web of things based knowledge graph mining

Recent developments in research have shown that knowledge graphs (KG) are successful in supplying useful external knowledge to enhance recommendation systems (RS). High-order connections between two items with one or more related qualities can be encoded in a knowledge graph. It is now feasible to extract both object properties and relations from KG with the aid of developing Graph Neural Networks (GNN), which is crucial for making good suggestions. In this study, word2vec creates virtual neighbors for nodes to make up for the lack of social connection data and knowledge graph representation learning mines item characteristics to extract additional data from the graph structure of social interactions. Numerous experimental findings on the two simple data sets, Douban and Yelp, show the accuracy and superiority of the MSAKR algorithm. These include a knowledge graph, multi-task training, training the recommendation module, and the knowledge graph concurrently to integrate item attributes. The information graph's deep reasoning capability is not utilized in this study; instead, ordinary representation learning is used. Social interactions support recommendations as well. The experimental findings demonstrate that the suggested model performs better than alternative benchmark models.