文章提出一种改进的关联规则方法,用于抽取文本中的非分类关系。首先利用基于上下文的术语相似度获取方法得到术语间的相似度权重,再通过加入谓语动词的关联规则算法计算,结合搜索引擎技术得到候选关系对集合,并通过置信度和支持度的对比分析,抽取最终的非分类关系结果,最后对测试数据进行实验,并对结果进行分析。 This paper presents an improved association rule method for extracting non-categorical relations in texts. First of all, we use the context-based similarity retrieval method to obtain similarity weight between terms, and then calculate the association rules by adding predicate verbs, and combine the search engine technology to get the candidate pairs. By comparing and analyzing the confidence and support, Extract the final non-classification results, and finally test the test data, and analyze the results.