A level-wise search approach to frequent web access pattern discovery with hybrid reasoning

Abstract

Discovering frequent patterns from ontologies has become a central topic in the Semantic Web. To find more efficient frequent web access patterns, the hybrid reasoning framework DatalogSHIQ(D) is defined to express log ontology knowledge base, in which DL language SHIQ(D) and restricted datalog rules are combined with the basic datalog safeness. After introducing the expression of web access pattern and the tasks to the mining problem, an ILP approach is illustrated to generate the candidate frequent web access pattern set by breadth-first expansion. The experimental results show that our approach can discover more expressive web usage information without increasing calculation complexity compared with previous work.