Aiming at the problem of lacking knowledge representation and semantic processing capacity in traditional information retrieval, this paper proposes an ontologybased semantic web information retrieval framework, which includes the knowledge collection and storage layer, the humancomputer interaction layer and the knowledge processing layer. And the realization of the key technologies in the knowledge collection and storage layer is described, including the construction of the domain ontology library, metadata extraction,and the construction of knowledge base. Among them, the realization of the core technology in metadata extraction, such as the domain ontology library analysis, document preprocessing, semantic annotation, knowledge extraction, etc. is described in detail.