The notion of ontology is introduced to information retrieval systems and semantic Web-oriented architecture of personalized intelligent information retrieval system is presented, and hierarchical tree model of local domain ontological repository of biomedical diseases system is constructed, in which the OWL language is used to detail annotate syntax and semantics of concepts, their attributes and relations among the concepts. Automatic acquirement of remote information on the Internet and pretreatment of local domain ontological repository are implemented by inference rules and information semantics-based intelligent clustering algorithm in the architecture, and preference database of doctors are managed logically according to their retrieval interests. Consequently, the system will meets the requirement of personalized intelligent information search for diseases diagnosis, and further improves recall and precision of information retrieval on biomedical diseases system.