基于TF-IDF和互信息的推荐算法研究

Research on recommendation algorithm based on TF-IDF and Mutual-information

本文提出一种基于TF-IDF和互信息的方剂推荐算法。其核心思想是根据TF-IDF算法的原理,确定核心药物;再计算核心药物和方剂间的互信息来确定二者相关性,以此确定最有效的方剂。对名老中医治疗肺癌的542首方剂,共计342味药物进行数据挖掘,通过该算法获得核心药物71味,推荐方剂126首。采用该算法获得名老中医治疗肺癌的核心方剂的结果表明,该算法通用性强,效率高。由于不仅探索了药物层面的规律,还挖掘了方剂层面的信息,故该算法有较高的实用价值。

This paper proposes a prescription recommendation algorithm based on TF-IDF(Term Frequency-Inverse Document Frequency)and Mutual-information.The core idea is to determine the core drug according to the principle of TF-IDF algorithm.Then,the Mutual-information between the core drug and the prescription is calculated to determine the correlation between the two,so as to determine the most effective prescription.Through the data mining of 542 prescriptions of TCM treatment for lung cancer,a total of 342 drugs,71 core drugs and 126 recommended prescriptions were obtained by this algorithm.The result of obtaining the core prescription of famous herbalist doctors in the treatment of lung cancer with this algorithm shows that the algorithm has strong universality and high efficiency.The algorithm is of high practical value because it not only explores the law of drug level,but also excavates the information of prescription level.