Abstract
High frequent network request pattern in office automation system (OAS) can be described as high frequent item set when association rule discovery technology is used to mine high frequent network request patterns with network log file as raw data. To mine high frequent network request concerned by user (or manager), attention set is used to describe the focus of user during the mining of high frequent network request. In this paper, extension method is discussed for the extension of attention set and extension method based on the co-occurrence of item pair (network request behavior pair) is presented. Experimental results show that the performance of our proposed method is better.