Human experience is a special kind of knowledge, which captures previously experienced, similar problem situations (case) for human problem solving. This paper describes a case-based approach for knowledge sharing and management with regard to the semantic web settings. We define a RDF-based Case Markup Language (CaseML) for experience knowledge representation and develop a generic architecture for the case based reasoning in the open, distributed environment with the emphasis on the integration of case knowledge with web ontologies. An application scenario from Traditional Chinese Medicine field is introduced to evaluate our methodology. With our application experience, we claim that, in contrast with logic-based knowledge representation formalisms, case is a more suitable for knowledge sharing in the semantic web, because case based reasoning is similarity-based and there is weak demand on semantic consistent in such reasoning processes.