Abstract: A semantic-based session analysis method is presented.Semantic Web usage log preparation model enhances usage logs with semantics.Markov chain model based on ontology semantic measurement is used to identify which active session a request should belong to.A competitive method is applied to determine the end of sessions.Results of practical application showed that the average percentage of successfully analyzed sessions by the semantic-based session analysis method is 69.8%,higher than other algorithms,such as the timeout and referrer-based method.