Mining Unexpected Web Usage Behaviors

Abstract : Recently, the applications ofWeb usage mining are more and more concentrated on finding valuable user behaviors from Web navigation record data, where the sequential pattern model has been well adapted. However with the growth of the explored user behaviors, the decision makers will be more and more interested in unexpected behaviors, but not only in those already confirmed. In this paper, we present our approach USER, that finds unexpected sequences and implication rules from sequential data with user defined beliefs, for mining unexpected behaviors from Web access logs. Our experiments with the belief bases constructed from explored user behaviors show that our approach is useful to extract unexpected behaviors for improving the Web site structures and user experiences.
Document type :
Conference papers
Complete list of metadatas

https://hal-lirmm.ccsd.cnrs.fr/lirmm-00275952
Contributor : Haoyuan Li <>
Submitted on : Tuesday, March 24, 2009 - 11:04:02 AM
Last modification on : Monday, February 11, 2019 - 6:22:02 PM
Long-term archiving on : Friday, May 28, 2010 - 5:48:02 PM

File

llp08incdm.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : lirmm-00275952, version 1

Collections

Citation

Haoyuan Li, Anne Laurent, Pascal Poncelet. Mining Unexpected Web Usage Behaviors. ICDM'08: 8th Industrial Conference on Data Mining, pp.283-297. ⟨lirmm-00275952⟩

Share

Metrics

Record views

156

Files downloads

920