Knowledge embedding towards the recommendation with sparse user-item interactions

• DOI: 10.1145/3341161.3342876
• Publication Date: 2020-01-15T21:07:04Z
• Authors: Deqing Yang; Ziyi Wang; Junyang Jiang; Yanghua Xiao
• Keywords: Word embedding; Representation; Feature Learning; Graph Embedding

Recently, many researchers in recommender systems have realized that encoding user-item interactions based on deep neural networks (DNNs) promotes collaborative-filtering (CF)'s performance. Nonetheless, those DNN-based models' performance is still limited when observed are very less because the training samples distilled from these critical for learning models. To address this problem, we resort to plenty features knowledge graphs (KGs), profile users and items precisely sufficiently rather than interactions. In paper, propose a embedding recommendation framework alleviate problem of sparse recommendation. our framework, each user item both represented by combination an tag at first. Specifically, embeddings learned Metapath2Vec which graph model qualified heterogeneous information networks. Tag Skip-gram similar word embedding. We regarded as they indicate about latent relationships movie-movie user-movie. At last, target user's representation candidate movie's fed into multi-layer perceptron output probability likes item. The can be further used achieve top-n extensive experiments movie dataset demonstrate framework's superiority over some state-of-the-art models, especially scenario user-movie