摘要
在知识图谱的知识表示学习中,绝大多数方法都是将知识图谱中实体和关系映射到低维的连续向量空间中,但数据的稀疏和信息的不足仍会导致实体和关系语义表示的不完整性。针对这个问题,提出基于图上下文的知识表示学习模型(Context_RL)。将知识图谱中实体和关系的上下文信息作为可靠数据来源和输入。通过计算实体和关系的语义相似度,将图上下文信息融入向量表示中,在语义的层面上提高了知识图谱表示学习的能力。通过实体预测和三元组分类实验,在有关数据集上,Context_RL的实验结果比其他模型更好。
In the knowledge representation learning of knowledge graphs,most methods map entities and relationships in the knowledge graphs to low-dimensional continuous vector space,but the sparseness of data and the lack of information will still lead to incompleteness of semantic representation of entities and relationships.A context-based knowledge representation learning model(Context_RL)is proposed to solve this problem.The context information of entities and relationships in the knowledge graph was taken as the reliable data source and input.By calculating the semantic similarity between entities and relationships,the graph context information was integrated into vector representation,which improves the ability of knowledge graph representation learning on the semantic level.Through the experiments of entity prediction and triple classification,the result of Context_RL is better than other models on the relevant data sets.
作者
周泽华
陈恒
李冠宇
Zhou Zehua;Chen Heng;Li Guanyu(College of Information Science and Technology,Dalian Maritime University,Dalian 116026,Liaoning,China;School of Software,Dalian University of Foreign Languages,Dalian 116044,Liaoning,China)
出处
《计算机应用与软件》
北大核心
2021年第6期120-125,共6页
Computer Applications and Software
基金
国家自然科学基金项目(61371090,61602076,61702072,61976032)
国家社会科学基金项目(15BYY 028)
辽宁省自然科学基金项目(20170540232,20170540144,20180540003)
大连外国语大学研究创新团队项目“计算语言学与人工智能创新团队”(2016CXTD06)。
关键词
知识图谱
知识表示学习
图上下文
语义相似度
向量表示
Knowledge graph
Knowledge representation learning
Graph context
Semantic similarity
Vector representation
