| A big flaw of Internet is that it doesn't provide enough semantic information about web resources. Computers cannot understand what the information mean. This is why WWW is transforming to semantic web. Resource Description Framework (RDF) is used to implementing semantic description of web resources.In this paper, we have done a lot of research on RDF inference mechanism and proposed a semantic retrieval framework based on RDF inference mechanism. This framework is composed of domain modeling of given domain, semantic description of information resources, extension search of domain model, and etc. It realizes semantic description of information resources to resolve the unresolved problem of identifying synonyms and relevant information retrieval in current information retrieval system.This paper emphasizes on RDF inference mechanism, which is the theoretic fundamental of semantic retrieval. First we build the model of application domain by RDFS ontology, and then add the RDF description about domain resources. The incorporation of these two forms the knowledge base, which can support limited inference according to RDF and RDFS entailment rules. But there are some limitations of RDF semantics. For example, RDF Schema can only organize the concepts according to hierarchy and can't specify some characteristics of properties (such as transitivity, symmetry and etc). In order to remedy the limitation, we dynamically extend the domain model through domain extension rules to provide semantic retrieval a sounder model.We have developed a RDF based semantic retrieval system on computer network field and expected to provide an intelligent retrieval service for learning resource. This system tries to solve problems that ordinary systems cannot do such as identify synonyms, relative information retrieval. The result of experimentation indicates that this retrieval framework makes semantic of resources sufficiently emerged and the the search strategy is feasible. |
