Study on web mining data source pattern based on an application server log
Dong Rong-he · Jisuanji gongcheng yu sheji · 2009
The quality of the web mining data source has directly affected the result of marketing and recommendation algorithm in E-commerce systems.Now there are a lot of E-commerce systems which use web logs as the web data source.They limit the functions of analysis and decision of E-commerce and may make mistakes.An application server log based pattern EUP is proposed.It can make the user session identification methods more accurate and efficient and can solve a series of preprocessing problems existing in web logs.It has integrated purchase,browse and other kinds of events.It is compatible with other web mining algorithms.It can support multi-dimension and multi-level web mining analysis.The experimental results show that web data quality is greatly enhanced,a good foun-dation for intelligent services is provided.