Database Reference
In-Depth Information
Figure 18: All signifi cant access paths
Online Analytical Mining of Path Traversal Patterns
In web usage, users activities on web sites are recorded into server log fi les continuously,
even though path traversal patterns have been derived before. As a result, the derived path
traversal patterns are outdated soon. To maintain the current status of the path traversal
pattern, we update the user access patterns continuously or periodically, whenever the log
fi le is being updated. This is accomplished by time scheduling.
Suppose the access log is being updated after a period according to the time set in the time
scheduling part. As a result, an up-to-date user accessed pattern has been maintained.
The system provided some Online Analytical Processing (OLAP) functions, including
roll up and drill down. The following fi gures show the up-to-date user access patterns and
statistics summary.
Example 2
The target web page 'd_resrch' and time dimension “Date” are selected and the
Confi dence and Support thresholds were set to 50% and 3% respectively. Then a set of access
patterns were generated if their confi dence and support levels were greater than or equal to
the values inputted by the user. Figure 18 displays the result of the query.
Example 3
The target web page 'd_resrch' and time dimension “Month” were selected and the
Confi dence and Support thresholds were set to 50% and 3% respectively. Then a set of access
patterns was generated if its Confi dence and Support levels were greater than or equal to the
values inputted by the user. Figure 19 displays the result of the query.
Besides the user navigation paths, useful statistics are also provided for analysis. By
clicking the button “Statistics”, a screen appears. Specifi cally, the main module of OLAM
provides four difference statistics. Executive summary provides a general statistic result for
