甲骨文知识图谱构建中的实体关系发现研究

Research on entity relation discovery for Oracle Bone Inscriptions knowledge mapping construction

为解决甲骨文信息处理研究中对专家的依赖性高而专家知识共享率却很低这一突出矛盾,提出构建知识图谱的方案,其核心问题是如何发现实体和实体间的关联关系。基于甲骨文的古籍特征,先通过知识图谱MKD的共词分析发现实体及其关系,再利用甲骨文本体和规则推理发现隐含的语义关系,最后将其与MKD共词关系进行融合。实验结果表明,该方法可以有效解决甲骨文知识图谱中实体关系发现问题。

During the research process of Oracle Bone Inscriptions （OBI） information processing, there is a prominent contradiction between high dependence on experts and low sharing of their OBI knowledge. In order to solve this problem, we propose an OBI knowledge mapping scheme, the crucial issue of which is how to find the ontology relations. Firstly,based on the OBI features,the co-word analysis of Mapping Knowledge Domains （MKD） is done to find the ontology relations. Secondly, OBI on- tology and reasoning are used to find the implicit semantic relations. Finally, the co-word relations and ontology relations are fused to describe the OBI knowledge mapping relations. Experimental results show that the method can effectively solve the problem of entity relation discovery in OBI knowledge mapping construction.