基于个人经历知识图谱的领域评审专家亲密度计算

Domain reviewer closeness calculation based on personal experience knowledge graph

针对领域评审专家与申请人之间亲密度过高影响评审结果公正性的问题,提出了面向个人经历知识图谱的亲密度计算方法。首先采用自顶向下与自底向上相结合的方法构建知识图谱;然后从中查找专家与申请人间建立的多条社会关系传播路径,并利用专家评价法对其量化处理;最后分析直接关系和间接关系的形成特点,建立亲密度计算模型。实验表明,该知识图谱可有效表示基于专家个人经历建立的关系网络,挖掘隐含的多条亲密社会关系,提高专家亲密度计算的准确性。

In order to solve the problem of overly close relationship between reviewers and project applicants affecting the fairness of review results,a closeness calculation method oriented to the personal experience knowledge graph was proposed.Firstly,the knowledge graph was constructed by the combination of top-down and bottom-up methods.Then,the multiple communication paths of social relations established between experts and applicants were found and quantified using expert evaluation method.Finally,the formation characteristics of direct and indirect relationship were analyzed,and the closeness calculation model was established.The experimental results show that the knowledge graph can effectively represent the relationship network established based on expert personal experience,uncover hidden multiple intimate social relationships,and improve the accuracy of reviewer closeness calculation.