Recommender Systems Based on Graph Embedding Techniques: A Review

As a pivotal tool to alleviate the information overload problem, recommender systems aim predict user’s preferred items from millions of candidates by analyzing observed user-item relations. for alleviating sparsity and cold start problems encountered systems, researchers generally resort employing side or knowledge in recommendation as strategy uncovering hidden (indirect) relations, aiming enrich (or data) recommendation. However, face high complexity large scale knowledge, this largely relies efficient implementation on scalability models. Not until after prevalence machine learning did graph embedding techniques be recent concentration, which can efficiently utilize complex large-scale data. In light that, equipping with has been widely studied these years, appearing outperform conventional implemented directly based topological analysis resolution). focus, article systematically retrospects embedding-based bipartite graphs, general graphs proposes design pipeline that. addition, comparing several representative models most common-used simulations, manifests that still overall ones predicting implicit interactions, revealing comparative weakness tasks. To foster future research, constructive suggestions making trade-off between different tasks, puts forward some open questions.