The Study Of XML Storage Technique In Relational Database

XML has already become the standard format for data present and data exchange onthe Internet. With the sharp increasing of XML data, many kinds of DBMS come out inuse to manage XML data, such as XML native database, object-oriented database,relational database and so on. Among them, use relational database to storage andmanage XML data is the dominant approach up to now or even for a long period of timeaccording to mature technique and the wide apply of relational database.Owing to different structure between XML schema and relational schema, weshould do schema mapping at first. There are two kinds of schema mapping methods.The first method is based on a set of fixed mapping rules,and by these rules we canproduce a relational schema for the XML documents. Another method is based on querycost. A XML schema can be mapped to many relational schemas, the cost-based schemamapping method is to search the best relational schema whose query cost is theminimum.This paper studies the storage of XML document in relational database based onquery cost. The paper introduces various storage approaches of XML document andpoints out advantage and disadvantage of these approaches at first. At the same time, thepaper puts forward an improved approach, and it has many characteristics as follows.First, the paper introduces an extended schema graph based on XML Schema. Theextend schema graph can straightforward and complete express the structure of XMLdocuments. Second, add '|'operate to handle choice elements or choice group elementsfor traditional operates can't to do under this situation. Third, when obtain neighboringschemas, map a simple type node to a single relation is limited, thus schema searchspace is smaller then before, which make search algorithm find the best schema inshorter time. Forth, a new query cost model based on the characteristics of hash joinalgorithm is put forward, this cost model is more accurate than others. Fifth, a greedyalgorithm is used as the search algorithm of schema space, and optimize the algorithmwhen compare the cost of two schemas. The optimized algorithm speed up the processof cost compare, and proved the performance of algorithm as a result.At last, I do the imitated experiment and obtained the outcome expected.