基于异构信息网络与TF-IDF的核心药物发现算法

Core drug discovery algorithm based on heterogeneous information network and TF-IDF

提出一种基于异构信息网络和TF-IDF的核心药物发现算法。其核心思想是建立包含症状、方剂等多种类型对象的异构信息网络,并使用PathSim算法得到方剂之间的相似度来完成方剂聚类。以此为基础使用综合了剂量因素与TD-IDF算法原理的药物重要性系数计算方法完成核心药物发现。本文从《伤寒论》的方剂中划分出9个主要聚类并给出了各个聚类上重要性排名前5的药物,该算法可以考虑到多方面的信息,合理地挖掘出核心药物。

In this paper,a core drug discovery algorithm based on heterogeneous information network and TF-IDF is proposed.The core idea is to set up heterogeneous information network including symptoms,prescriptions and other types of objects,and use PathSim algorithm to get the similarity between prescriptions to complete the clustering of prescriptions.Based on this,the core drug discovery is completed by using a drug importance coefficient calculation method that integrates the dose factor and the principle of TD-IDF algorithm.Nine main clusters are divided from the prescriptions of"Treatise on Febrile Diseases"and the top five drugs in each cluster are given.The algorithm can take into account many kinds of information to mine the core drugs reasonably.