A Study And Implementation Of Ontology Mapping Technology Based On Semantic Association

In recent years, with the rapid development of knowledge engineering, artificial intelligence, and information retrieval, Sematic Web has become an important foundation for collaborative work between different computers, and many applications have been turning from the stage of data processing to knowledge processing. As the knowledge representing form for Semantic Web, ontology has been widely used to express the semantic relationships between knowledge. However, the different understanding on same concepts from different domains by various organizations or persons has leaded the existence of heterogeneity between different ontologies, which make it difficult for data exchanging and semantic converting. Based on these problems, this paper studies how to create mapping relations between database schema and ontology model, and how to create mapping relations between different ontologies. Major contributions in this paper are outlined as follows:1. Propose an ontology mapping algorithm based on semantic association to achieve the creation of mapping relations between different domain ontologies. Firstly, by the study of semantic similarity computing, the preliminary mapping relationships between ontologies to be mapped have been created. Secondly, to improve the accuracy of mapping results, the semantic information contained in history mapping data has been involved in the process of ontology mapping. Lastly, the test sets of Word Similarity-353 and Words-240 have been used to measure the accuracy of mapping results. And it indicates that the accuracy of this algorithm can reach 80%, and the accuracy has also been improved to a different degree after the algorithm is optimized by history mapping information.2. Propose a semantic mapping algorithm between relational schema and ontology model based on pattern matching to achieve the mapping and transformation between relational schema and ontology model. Firstly, the relational schema and ontology model should be parsed and converted into the form of graph. Secondly, by using the intermediate models as inputs, the mapping results are created by the pattern matching algorithm and the similarity iterative algorithm. Lastly, the test cases provided by the Map Onto project developed by Toronto University are used to measure the mapping results by this algorithm. And it proves that for some complicated test cases, the experimental results are not quite ideal, but for other test cases, the algorithm can basically meet the experimental requirements.3. A cross-domain data semantic mapping system has been developed, which aims at creating mapping relations between different domain ontologies and between relational schemas and ontology models. For the requirements of building semantic mapping relations between heterogeneous data sources, the system has been developed, which verifies the feasibility and effectiveness of the algorithms proposed in this paper and meets the requirements of semantic sharing for cross domains. And it also proves that the system can meet the demands for increasing interoperability of integrated electronic information systems, which reaches the predetermined research objectives.