Affiliations: [a] Department of Computation, UMIST, PO Box 88, Manchester, M60 1QD, UK | [b] Department of Computer Science, University of Westminster, Northwick Watford Rd, Northwick Park, London, HA1 3TP, UK
Abstract: Web log analysis can provide valuable information about user behaviour when visiting a web site. Analysing users' traversal patterns and frequency or repetition of visits can provide very valuable information. However, the amount of information contained in web logs is huge. Primarily based on the Web servers' log files, the main objective of traversal pattern mining is to discover the frequent patterns in users' browsing paths and behaviours. Constraints have not been considered extensively as part of pre-mining tasks in web mining. This paper presents a complete framework for web usage mining, allowing users to pre-define constraints when analysing complex traversal patterns in order to improve the efficiency of algorithms and offer flexibility in producing results.
