一种Web事务识别的新模型及其频繁路径挖掘

针对已有Web事务识别模型的缺点,提出一种识别Web事务的新模型———IPRC模型.该模型根据主索引页上的引用以及文档目录结构将网页分类,并以此作为识别Web事务的依据.在此基础上提出了一种挖掘频繁访问模式的算法WDHP,该算法继承了DHP算法使用hash树过滤候选集以及裁剪数据库的基本方法,并以访问路径树的方式将数据库存储于内存,在内存中完成后继的挖掘,不仅减少了扫描数据库的次数,而且大大降低了算法的时间复杂性.实验表明WDHP算法不仅优于DHP算法,而且也优于典型的基于内存的WAP算法.

A new model for identifying internet transactions and mining frequent access paths mining

Existing models for internet transaction identification are limited.A new model is proposed that defines page categories according to the links on an index page and the structure of a file directory of a site.It also groups internet log data into web transactions according to page categories.Based on this, a new algorithm for frequent access paths mining is also proposed.The new algorithm inherits from direct hashing and pruning(DHP) the advantages of using a hash table to filter the candidate set and trimming the database.It puts the database into the main memory using a tree and finds frequent patterns on the tree. The algorithm needs less database scanning and significantly improves its time complicity.Experiments show that the algorithm outperforms the DHP algorithm and web access pattern(WAP),which is based on main memory in execution efficiency.