Incremental and interactive mining of web traversal patterns
Information Sciences: an International Journal
| Hi-index | 0.00 |
Mining frequent traversal patterns is to discover theconsecutive reference paths traversed by a sufficientnumber of users from web logs. The previousapproaches for mining frequent traversal patterns needto repeatedly scan the traversal paths and take a largeamount of computation time to find frequent traversalpatterns. However, the discovered frequent traversalpatterns may become invalid or inappropriate whenthe databases are updated. In this paper, we proposean incremental updating technique to maintain thediscovered frequent traversal patterns when the usersequences are inserted into or the database. Ourapproach partitions the database into some segmentsand scans the database segment by segment. For eachsegment scan, the candidate traversal sequences thatcannot be frequent traversal sequences can be prunedand the frequent traversal sequences can be found outearlier. Besides, the number of database scans can besignificantly reduced because some information can becomputed by our approach. The experimental resultsshow that our algorithms are more efficient than otheralgorithms for the maintenance of mining frequenttraversal patterns.