International Business Machines Corporation
Pattern-optimized session logs for improved web analytics

Abstract:

A Web-analytics system analyzes user activity on a Web site of interest. The system receives a set of user-session logs that each track user behavior on the site during one session. The system identifies common patterns of user behavior recorded in the logs. The system stores each identified pattern in a reserved area and replaces each instance of a stored pattern in a log with a placeholder. During the analytics system's subsequent analysis of a session, the system need process only the reduced-size log of that session, referring to the previously stored patterns whenever encountering a corresponding placeholder.