Abstract: Association rules analysis is an important technique of data mining techniques.FP-growth algorithm and MAXFP-Miner algorithm are representative algorithms.Name entities contain main content of the text，include abundant kowledge patterns.Based on the characteristic of the name entities，put forward a modify strategy of vector similarity comparison and applied it to MAXFP-Miner algorithm and gain animproved MAXFP-Miner algorithm.Applied this algorithm to analyze the potential associations the name entities，discover significant knowledge patterns.The result of experiment shows the improved MAXFP-Miner algorithm is effective and better than traditional FP-growth algorithm and MAXFP-Miner algorithm.

Key words: name entity, association rules analysis, MAXFP-Miner, knowledge discovery

摘要： 关联分析是数据挖掘技术中的一种重要方法，代表性算法有FP-growth算法和MAXFP-Miner算法。命名实体包含了文本的主要内容，蕴含了丰富的知识模式。针对命名实体的特点，提出一种基于向量相似度比较的关联规则修正策略，将此修正策略应用于MAXFP-Miner算法中，得到一种改进的MAXFP-Miner算法；利用该算法对命名实体之间的内在联系进行分析，从中发现有意义的知识模式。实验结果与性能比较表明，改进的MAXFP-Miner算法是有效的，优于传统的FP-growth算法和MAXFP-Miner算法。

关键词: 命名实体, 关联分析, MAXFP-Miner, 知识发现