Mu-En: Multi-path of Entity Recommendation Based on Path Similarity

摘要

In the process of exploring the path of similarity in entity recommendation, the selection of nodes in the path selection process attracts more and more attention. However, single-node path composed by one attribute is valued as the path in the face of the entity with multiple attributes for current research work, with adopting the path in similarity and advanced algorithm. In this work, we investigate the entity attributes in recommender system based on the link prediction method methods; on the basis of path similarity, we analyze the specific single-node paths, and also design the multi-node path composed of multi attributes. Therefore, this study not only provides more profound recommender results based on path similarity, but also greatly widens the path selection and improves the recommender accuracy.