aaai2016.pdf

Representation Learning of Knowledge Graphs with Entity Descriptions Ruobing Xie, Zhiyuan Liu, Jia Jia, Huanbo Luan, Maosong Sun Department of Computer Science and Technology, State Key Lab on Intelligent Technology and Systems, National Lab for Information Science and Technology, Tsinghua University, Beijing, China Jiangsu Collaborative Innovation Center for Language Ability, Jiangsu Normal University, Xuzhou 221009 China Abstract (embeddings). As a supplement to symbolic representation, the embeddings in latent space can significantly promote Representation learning (RL) of knowledge graphs aim- knowledge acquisition and inference (Bordes et al. 2013; s to project both entities and relations into a continuous low-dimensional space. Most methods concentrate on Yang et al. 2014; Neelakantan, Roth, and McCallum 2015). learning representations with knowledge triples indicat- Most existing RL methods solely learn from fact triples of ing relations between entities. In fact, in most knowl- KGs (Bordes et al. 2013). In fact, in most KGs there are also edge graphs there are usually concise descriptions for concise descriptions for entities, with rich semantic infor- entities, which cannot be well utilized by existing meth- mation about these entities. For example, in Fig. 1 we show ods. In this paper, we propose a novel RL method for