4.txt

This complete matlab for neural network

发信人: yaomc (白头翁&山东大汉), 信区: DataMining
标  题: [合集]请问web挖掘问题
发信站: 南京大学小百合站 (Wed Jan 16 17:06:21 2002), 站内信件

helloboy (hello) 于Wed Dec 26 11:13:16 2001提到：

I know preprocessing mainly divided into following parts:
1 data cleaning
2 user identification
3 session identification

but none of these 3 parts have common authoritive algorithms,
which makes result of preproceessing diffrenet.
I 'd like to know which algorithm are most authoritive?


roamingo (漫步鸥) 于Wed Dec 26 22:02:56 2001提到：

I don't know either.  But I think you can:
1). Take raw log file preprocessing as your field and focus 
    your research on it, or (probabaly more preferably)
2). Assume that you have perfect session files and build your
    mining algorithm on top of them, or
3). If you are going to build a web usage mining system, tackle 
    both, but without going down too deep into them.