Research Of Ontology-based Heterogeneous Data Integration

Under the background of intelligent city, with the continuous development of modern science and technology progress, all enterprises and institutions have established their own information system, realized the data informatization and network management. But the difference between enterprises makes the information systems are established in different ways by different developers and also in different period. The developers only consider of the business needs of their own system, leading to the difference on system platform, database technology, data structure and query language. The massive data are stored in various forms, and depends on different database management system. They are called heterogeneous data. These heterogeneous data are structural heterogeneous, distribute and independent. That makes the data between various information systems cannot share with each other, formed the "Information Island". Therefore, how to realize the heterogeneous data integration is particularly important.Many heterogeneous data integration solutions have been proposed at home and abroad, some heterogeneous problems can be solved effectively, especially the grammar heterogeneous. But the semantic heterogeneous problem has not been solved very well. This paper introduces the concept of ontology to solve the problems of semantic heterogeneity of heterogeneous data integration.This article first expounds the concept of heterogeneous data and the goal of heterogeneous data integration, summarizes several typical heterogeneous data integration system structure and its advantages and disadvantages. After introducing the concept of ontology and summarizing the ontology-based heterogeneous data integration methods and its advantages. The whole heterogeneous data integration system frame structure is given, and the key problems in the process of integration are discussed.Secondly, this paper does research on the ontology mapping. After studying the existing ontology concept similarity algorithm, this paper proposes a modified way of integrated domain ontology similarity algorithm. The modified algorithm firstly searches the existence of concepts in world knowledge system, like WordNet and HowNet, to avoid the field limitations of the concept similarity algorithm. At the same time, the modified algorithm combines semantic similarity, attribute similarity and structure similarity algorithm to calculate the integrated similarity, avoiding the one-sided calculation process, to reach the purpose of improving accuracy of domain ontology similarity. That lay a solid foundation for the mapping between ontology and subsequent query expansion.Finally, the modified integrated domain ontology similarity algorithm is applied to individual utilities accounts management platform, the actual heterogeneous data system. Compared by the traditional single concept similarity algorithm, the modified algorithm has high matching success rate and higher matching accuracy, and its comprehensive matching rate is better than a single algorithm. This modified integrated domain ontology similarity algorithm can integrate the heterogeneous data which are at the bottom of heterogeneous application system, can realize the function of querying the individual utilities bills on "one-stop" unified platform, embody the application value of the modified algorithm.