Data Integration, Enterprise Application System Uncertainty

As the rapid development of the network technology and the computer technology, the enterprise competition is becoming increasingly fierce. The formation of information islands forced the Enterprise Application Integration have to be carried out within enterprise or between enterprises dynamically. At present, the enterprise application system integration have various forms, this paper mainly discussed the data integration. However, in traditional database applications, the traditional data management technology didn't consider the existing of data uncertainties in reality. Many existing methods eliminated the uncertainties with choosing the results which most likely matched users' intents. But sometimes these uncertain information is of value, some useful information may not be returned to the users if ignore this information. In order to solve this problem, this paper mainly discussed the processing ways of uncertain data in Data Integration Systems.For all the inconsistencies problems of the data in the source database, the third chapter of this paper put forward a solution method based on the fuzzy multiple attribute. Users can get certain search results by using this method. First, we should get the quality indexes according to certain methods; Second, Whether these indexes are qualitative or quantitative, we normalized them by using triangular fuzzy function; Third, constructed a fusion matrix; Fourth, blurred based on this fusion matrix; Fifth, calculated the distance between data sources si and positive/ minus ideal solutions, then calculated the degree of membership of si is an ideal solution; Sixth, chose the maximum degree of membership, then the data in si is the solving results. Finally, this paper inserted the resolve result into the search results set.For all uncertainties appears in all levels of the enterprise application systems, the fourth chapter of this paper put forward a data integration method based on probability database and probability model mapping. This paper uses probability database to solve the uncertainties in data source level, and uses the probability model mapping methods to solve the uncertainties in model mappings and queries. To solve the uncertainties in data source level, system designer build probability database and the system administrator fill in possible data with different probabilities. For the uncertainties in model mappings and queries, we can give each model mapping a probability value according to the investigation of users'query request or the query conditions and users'real intentions by using the automation tool, namely the corresponding relation between the users'query and target query. When a user input query conditions, the system decide the target queries based on the probability value. Finally it returned to users the final query results. These results can be all possible query results, and also can be top-k optimal query results according to users'requirements. Finally, this paper verified the feasibility and correctness of these two methods by the system of "The integration in CRM of Zhong Tai Yang Guang".