基于社区发现算法的消渴六经证候研究＊

目的 通过社区发现算法挖掘消渴病六经证候分布规律。方法 在已构建的消渴病经方古籍知识图谱基础上，对图结构数据进行症状及其关联的提取，运用Louvain算法进行社区发现，运用PageRank算法分析各社区中症状的主次。另采用因子分析方法进行证候挖掘，并对二者结果进行比较。结果 由社区发现得到4个社区，其六经证候分别为阳明病、阳明太阴合病、太阴病（水证）、太阴病（血证）。由因子分析得到10个公因子，六经证候为阳明病（X2、X4、X5、X7、X8），太阴病（X1、X9、X10）、阳明太阴合病（X3）。结论 基于社区发现的证候研究方法，可有效挖掘古籍中消渴病的六经证候规律，其结果在专业上具备较强的可解释性。

A Study on Six Meridians Syndrome of Diabetes (Xiaoke) Based on Community Detection Algorithm

Objective To explore the distribution pattern of six meridians syndrome of diabetes (Xiaoke) by community detection algorithm.Methods On the basis of constructing a knowledge graph of the ancient books on the classical formulas for diabetes (Xiaoke), the symptoms and their connections were extracted from the data of graph structure. Community detection was carried out by Louvain algorithm, and the importance of symptoms in each community was analyzed by PageRank algorithm. Separately, the distribution pattern of six meridians syndrome was explored by factor analysis. The explore results were compared.Results Four communities were found by community detection algorithm, and their six meridian syndromes respectively were Yangming Disease, Yangming-Taiyang Joined Disease, Taiyin Disease (water syndrome), Taiyin Disease (blood syndrome). Factor analysis obtained 10 categories of factors, their six meridians syndromes were Yangming Disease (X2, X4, X5, X7, X8), Taiyin Disease (X1, X9, X10), Yangming-Taiyang Joined Disease (X3).Conclusion The syndrome research method based on community detection can effectively excavate the rules of six meridian syndrome distribution in the ancient books of diabetes (Xiaoke), which is more interpretable in professional field.