运用哈希技术进行关联规则挖掘

以前基于支持度一置信度框架的关联规则挖掘算法都是先用支持度做为阈值对搜索结果进行剪枝 ,产生频繁集 ,再针对频繁集产生关联规则 ,这就是频繁关联规则。然而在很多应用中 ,诸如 :鉴别相似的Web文件、网络中入侵检测等 ,有许多有趣的关联规则仅有很少的支持度。在本文中 ,针对这种情况 ,提出了一种可以挖掘非频繁项之间有趣规则的算法 ,此算法先用相似度作为兴趣度度量对算法结果进行剪枝

Association rule mining with the technology of hashing

All algorithms of support confidence framework based association rule mining firstly prune search results with the threshold of support. They all product frequent association rules. Whereas, in many applications such as identifying of similar web documents, detecting the intrusion in Internet, clustering, etc. In these tasks, there are many interest rules that have comparatively few instances in the data set. In this case, We must look for highly correlated items, which may be infrequent items. In this paper, I propose a kind of technology or algorithm that can mine rules between infrequent items, which uses the similarity threshold to prune the search results first.