基于TF-IDF相对熵的中医证候量化研究

提出了用术语频率-逆文档频率（Term Frequency - Inverse Document Frequency，TF-IDF）相对熵作为证候量化的表示方法。TF-IDF思想来源于文本信息挖掘，是文本自动分类中一种重要的方法。TF-IDF算法也体现了中医证候的自动分类思想：一个症状在特定证候中出现的频率越高，说明它在区分该证候方面的能力（即TF）越强；一个症状在所有证候中出现的范围越广，说明它区分某证候的能力（即IDF）越低，并用具体实例进行了验证。

Quantitative Research on Traditional Chinese Medicine Syndrome Based on TF-IDF Relative Entropy

This study proposed to use Term Frequency - Inverse Document Frequency (TF-IDF) relative entropy as knowledge representation method between symptoms and syndrome. TF-IDF was originated from text mining. It was an important method in the automatic text categorization. TF-IDF also represented the automatic categorization idea in traditional Chinese medicine (TCM) syndrome. It was based on the fact that the higher frequency of one symptom in specific syndrome, the stronger ability to distinguish this syndrome (TF); and the more wide range of one symptom in all syndrome, and the lower ability to distinguish a syndrome (IDF). It was verified with specific examples.