Embodying the Number of an Entity's Relations for Knowledge Representation Learning.
Xinhua Suo,Bing Guo,Yan Shen,Wei Wang,Yaosen Chen,Zhen Zhang
DOI: https://doi.org/10.1142/s0218194021500509
IF: 1.007
2021-01-01
International Journal of Software Engineering and Knowledge Engineering
Abstract:Knowledge representation learning (knowledge graph embedding) plays a critical role in the application of knowledge graph construction. The multi-source information knowledge representation learning, which is one class of the most promising knowledge representation learning at present, mainly focuses on learning a large number of useful additional information of entities and relations in the knowledge graph into their embeddings, such as the text description information, entity type information, visual information, graph structure information, etc. However, there is a kind of simple but very common information - the number of an entity's relations which means the number of an entity's semantic types has been ignored. This work proposes a multi-source knowledge representation learning model KRL-NER, which embodies information of the number of an entity's relations between entities into the entities' embeddings through the attention mechanism. Specifically, first of all, we design and construct a submodel of the KRL-NER LearnNER which learns an embedding including the information on the number of an entity's relations; then, we obtain a new embedding by exerting attention onto the embedding learned by the models such as TransE with this embedding; finally, we translate based onto the new embedding. Experiments, such as related tasks on knowledge graph: entity prediction, entity prediction under different relation types, and triple classification, are carried out to verify our model. The results show that our model is effective on the large-scale knowledge graphs, e.g. FB15K.