Computer Engineering and Applications ›› 2022, Vol. 58 ›› Issue (21): 294-300.DOI: 10.3778/j.issn.1002-8331.2104-0286

• Engineering and Applications • Previous Articles     Next Articles

Research on Quantitative Evaluation of Knowledge Fusion in News Knowledge Graph

XIE Tianyang, CHEN Ming, XI Xiaotao   

  1. 1.College of Information Technology, Shanghai Ocean University, Shanghai 201306, China
    2.Key Laboratory of Fisheries Information, Ministry of Agriculture and Rural Affairs, Shanghai 201306, China
  • Online:2022-11-01 Published:2022-11-01

新闻知识图谱中知识融合量化评估研究

解天扬,陈明,席晓桃   

  1. 1.上海海洋大学 信息学院,上海 201306
    2.农业部渔业信息重点实验室,上海 201306

Abstract: Most methods to judge the completeness and consistency of knowledge graph are manual methods, which are based on syntax checking and instance checking, or reasoning the logical rules in the knowledge base. But there is no specific method to judge the degree of fusion of knowledge graph. This paper presents a quantitative evaluation method based on power law, and the fusion results of same topic news knowledge graph are verified. It is found that with the increase of keyword node similarity, the degree of integration of nodes increases, the number of nodes after fusion drops sharply, the power law model fits good, and the adjusted coefficient of determination are all greater than 0.98. This test method can provide a new standard for testing the fusion degree of same topic news knowledge graph.

Key words: knowledge graph, knowledge fusion, semantic similarity, power law

摘要: 现阶段检验知识图谱融合程度的方法大多是基于句法检查和实例检验的人工方法,或对知识库中的逻辑规则进行推理,以此判断图谱是否具有完备性和一致性,但缺少对知识融合量化评估的具体方法。提出了一种基于幂律定律的量化评估方法,并用融合后的长江大保护新闻知识图谱节点数进行验证,发现随着节点相似度的升高,当前图谱节点的融合度增加,融合后节点数急剧下降,幂律模型拟合优度良好,调整决定系数均大于0.98。这一检验方法可为新闻知识图谱的融合度检验提供新的标准。

关键词: 知识图谱, 知识融合, 语义相似度, 幂律定律