Abstract: Traditional Chinese medicine formula (TCMF) is a main treatment measure in traditional Chinese medicine (TCM). However, a formula always contains many herbs and only some herbs play an important role in the formula, so it’s meaningful to find the core herbs and their combination rules. To solve the problem, this paper proposes an algorithm based on effect degree (ED) which includes three main steps. The three steps are core herbs discovery algorithm based on ED, network construct algorithm based on pointwise mutual information with herb pair ED (PMIED) and herb combination rules detection algorithm based on overlapping communities. Through experiments, this paper finds 42 core herbs and 30 kinds of herb groups in consumptive lung disease formulas. By analysis, all the 42 core herbs have a good effect on treatment of consumptive lung disease. Meanwhile 26 groups meet the herb combination rules and have a good effect on treatment of consumptive lung disease.

Key words: traditional Chinese medicine formula (TCMF), core herbs, herb combination rules, effect degree (ED), pointwise mutual information with herb pair ED (PMIED), overlapping communities

摘要： 中药方剂（traditional Chinese medicine formula，TCMF）是中医治疗的一种主要手段。然而一首方剂往往包含多种草药，这其中只有几种草药对治疗特定的病症起重要作用。因此，找出方剂中的核心药物和其配伍规律对研究中药方剂有非常重要的意义。针对该问题，提出了一种基于效用度（effect degree，ED）的核心药物及配伍规律发现方法。该方法包含三个主要步骤，分别是基于药物效用度的核心药物发现算法、基于带药对效用度的点式互信息（pointwise mutual information with herb pair ED，PMIED）的药物组网算法、基于重叠社团的高效药物配伍规律发现算法。通过实验，发现了肺痿方剂的42种核心药物和30种药物配伍，经分析和中医专家确认，42种核心药物对肺痿确有良好疗效，30组药物配伍中有26组符合药物配伍关系且对肺痿有良好疗效。

关键词: 中药方剂（TCMF）, 核心药物, 药物配伍规律, 效用度（ED）, 带效用度的点式互信息（PMIED）, 重叠社团