Database Reference
In-Depth Information
Figure 4
:
General architecture of OLAM on user access patterns
Overview
We have developed a general architecture for OLAM of Path Traversal Pattern on web
usage (Fong, Wong & Fong, 2000a, 2000b). The architecture divides the web usage mining
process into two main parts. The fi rst part includes the processes of transforming the web
data into suitable transaction form. This includes preprocessing, user identifi cation, session
identifi cation and data integration components. The second part includes the generic data
mining and pattern matching techniques such as the discovery of path traversal patterns as
part of the system's online analytical mining engine. The overall architecture for the web
usage mining process is depicted in Figure 4.
Firstly, the data collected from the web log goes through two steps. In the fi rst step
of data preprocessing, data loading and cleansing, the data is fi ltered to remove irrelevant
information (i.e., server request failures, authentication failures, etc.). All entries of the log
Table 1: Services provided by the system
Services
Explanations
Executive summary
General statistics results for the entire time period of the log data.
Path traversal patterns
To mine web user navigation paths to fi nd patterns in the user behav-
ior when traversing a web site.
Requested page summary
Pages access summary such as the most and least frequently re-
quested pages by visitors of a web site.
Date/time summary
Pages access statistics information of the total number of pages
viewed for the month, week and day time-intervals.
Entry/exit summary
Pages access statistics information of the entry and exit pages viewed
by visitors of a web site.
