Mining Web Sequential Patterns Incrementally with Revised PLWAP Tree

Since point and click at web pages generate continuous data stream, which flow into web log data, old patterns may be stale and need to be updated. Algorithms for mining web sequential patterns from scratch include WAP, PLWAP and apriori-based GSP. An incremental technique for updating already mined patterns when database changes, which is based on an efficient sequential mining technique like the PLWAP is needed. This paper proposes an algorithm, Re-PL4UP, which uses the PLWAP tree structure to incrementally update web sequential patterns. Re-PL4UP scans only the new changes to the database, revises the old PLWAP tree to accommodate previous small items that have become large and previous large items that have become small in the updated database without the need to scan the old database. The approach leads to improved performance.

• Type: Conference
• Year: 2004
• Where: WAIM
• Authors: Christie I. Ezeife, Min Chen
• Keywords: Information Management, PLWAP Tree, Sequential Mining Technique, WAIM 2004, Web Sequential Patterns