Research Of Data Model Evolution Method In The User-oriented Data View

Due to the increasingly furious competition in the market, the diverse enterprisebusiness demands improve the frequency of the enterprise database upgrade. Thus, itis necessary to modify data model frequently. In traditional database, the data modelshould be suspended when it be upgraded and the data migration is also carried out.Meanwhile, the application software, which depends on the database, needs toupgrade.It hard to satisfy the customers’ frequent requirements to the data model.Therefore, in order to solve all these problems, this thesis gives data modelevolution method in the user-oriented data view. This method gets new datarequirements from the user-oriented data view to rebuild the user data model andmatches with global data model and finds out the differences between them. ThenMSCS(Model Static Conversion Strategy) is used to calculate the operation sequenceof data model. Finally, it carries on global data model evolution operation to get theoptimizational global data model.The main issues of the thesis are as follows：1.It introduces the pattern matching algorithm based on data model of semanticdescription of data element.On the basis of achieving unification structure description of user data model andglobal data model, this thesis proposes data model matching algorithm which basedon data element semantic description. The algorithm, through beginning with thesemantic similarity of element nodes and structural similarity in the semantic tree,gets the mapping relationship between models by the weighted average. Thealgorithm regards data element as matching element, and solves the problem of lowpattern matching result rate through resulting from nonstandard user data view.2. It researched the MSCS (Model Static Conversion Strategy).Firstly, according to the operation granularity, this thesis categorizes evolution,and according to the mapping relationship between the solution of the conflict inmechanism, researches the Model static conversion strategy, To follow the evolutionrules of the operating entities and attributes,it comes out the sequence of evolution.Finally,evolution of global data model is achieved.3. It researched confliction and rules of evolution during model evolution.In view of the conflict,it has carried on the classification and induction in theevolution process of global data model, and puts forward the concrete resolution ofconflict about the model evolution. Thus,semantic or structural confliction is avoidedand feasible global data model after evolution is guaranteed.All of the research contents, this thesis puts forward research strategy of the datamodel evolution method in the user-oriented data view, and ensures that the research gets very good practical value.