摘要： 针对当前方剂检索服务大多只提供简单查询功能，不能满足方剂知识整理与累积的需求，提出方剂4个方面的相似度模型。对于组成药、功效和主治的相似性，引入集合相似度模型。对于名称相似性，引入基于字符串的相似度模型，并进一步提出基于统计和词典的相似度模型。通过分析比较各模型，挖掘出方剂间的关联，完成方剂聚类。

关键词: 相似度, 聚类, 统计

Abstract: Aiming at prescription context services only provides easy inquiry function nowadays, and it is insufficient to meet the requirements of carding and accumulation, this paper proposes similarity model for four aspects of prescription. It introduces set similarity model in prescription’s ingredients, efficacies and functions. String-based similarity model is brought in a model based on statistics and dictionaries is proposed. It excavates inherent correlations among prescriptions, and finishes cluster prescriptions.

Key words: similarity, cluster, statistics

中图分类号: 

• N945.12