Evaluation of Cold Syndrome Types based on Cosine Similarity and its application in intelligent diagnosis of TCM

In this paper, an automatic evaluation method of Cold Syndrome Types is proposed based on the method of Cosine Similarity. The method consists of the creation of a standard cold case database, the separation of input cold cases, synonym replacement and quadruple structured processing, quadruple weight assignment and similarity calculation. On the basis of Cold Syndrome Types evaluation, we further propose a learning and elimination mechanism for the standard cold case database, which improves the computing efficiency of the algorithm to a certain extent. It is showed in the experiments that the method can be used to initially assess the Cold Syndrome Types through the four aspects of Observing, Smelling, Consulting and Pulse-talking, which is of high value and significance as an auxiliary support for intelligent TCM diagnosis.