Research And Implementation Of XML Keyword Search Based On Semantic Relevance

With the development of the network technology, large amount of data come forth exponentially in the web. XML is becoming the standard to describe and exchange information on the Internet. XML is widely used in e-commerce, information systems, and digital libraries and so on. Naturally, efficient information retrieval from these great amounts of XML documents is becoming extremely important.XML keyword search becomes a research hotspot in XML data searching field in recent years. Compared with XQuery and other XML query language, XML keyword search has its unique advantages. The customer doesn't need to study complicated query language, nor need to have thorough understanding of the structure of the XML document. In fact, the customer only needs to supply the keywords related to the contents he is interested in, then the result can be returned. Furthermore, because the granularity of XML keyword search is based on elements, it can only return the parts of the document including a keyword so that the search is efficient.The main researchful contents of the thesis are as follows: Comparing entity-node, attribute-node of the E-R model with XML's, we defined Entity Sub Tree as an element of Semantically Relevant Unit. We defined the semantic relevance of keyword matching nodes which makes each keyword be related to its corresponding context semantics. This allows users to choose particular semantics which makes the result match users' real intentions for querying. We divided keywords into primary keywords and lesser keywords to improve query efficiency. Finding semantically relevant units improves the semantic relevance of the result. Based on the above, we described and implemented XML keyword search arithmetic based on semantic relevance. Because relational database has strict theory and mature technology, we use it to store the XML information. The result of our experiments demonstrates the effectiveness and the efficency of the new query method, and it does better in expressing the user's querying intention.