A Semantic-Based Approach To Translating Relational Data To XML Data

With the development of Internet technology, new applications such as E-Business are emerging rapidly, and the requirement for disposing information in depth is growing. Due to its connatural advantages, XML is used more and more widely for data representation and exchange. On the other hand, many data have been stored in relational database, and the case will be so for quite a long time. Often relational database are used to store and manage data and XML data are used as information exchange medium. Thus, the integration and information sharing between relational data and XML data becomes a significant research topic.On the basis of examing the fundamental features of XML and its applications on web, this paper tackles the problem of how to translate relational data into XML data while preserving semantic relationships among the data. First, we use simple element correspondences to describe how XML schema associates with relational schema. With the concept of semantic-tree, we analyse the semantic of the schemas. Second, we apply the correspondences of these two schemas to semantic-tree by SMOC (Semantic Mapping of Correspondence) algorithm, which converts the correspondences into a set of more complex mappings from source schema to target schema. These mappings are no longer simple element correspondences. They capture the semantic information of schemas, such as foreign-key constraints and nested structure. After that, we present a method to create new values for those in target schema, which haven't been assigned any correspondence from source schema. Finally, the correspondences are serialized by correspondence-priority graph based DTOC (Data Translation of Correspondence) algorithm. This solves the problem of how to determine the operation sequences during data translation. Besides, we propose an approach to translating semantic mappings into SQL, so that we can get relational data from the source. These data are put into the right position in the XML documents after searching the documents in a deep-first way by a matching algorithm, to accomplish the translation from relational data to XML data.A prototype named as Redtox is designed as well to implement the above ideas and methods, which is descrbed briefly in this paper. We believe that what we have done is significant to the related research.