Design And Implementation Of PBmice Based On J2EE-SSH Framework-PBMICE-SR Semantic Retrieval

With the purpose of expanding the application of PBmice, This paper proposes a semantic retrieval framework which called PBMICE-SR. The framework provides intelligent access to obtain data. It makes the filter work automatically execute, which once need a large number of professionals to analyze, it is so easy to make mistake.The paper is mainly about how to realize the PBMICE-SR framework based on the J2EE-SSH, which includes the research of the framework kernel technology, design and implementation of each module and the test analysis of the framework. The specific design of PBMICE-SR is divided into three stages.First, the retrieval data preparation, in which there are two core modules, including automatic extraction of biological meta-data and the unified data marking with RDF. In this part, using the theory of multi-classification support vector machine and weight calculation model, this paper proposes DAGSVM-BM25 model, a biological meta-data auto extraction model. And it use the SSH framework and Weka platform to implement the model. In the model test, it shows that the extraction system played a superior performance on the biological data, which average recall rate and precision are above 96%.Second, the data loading stage, in which the core function is to load the prepared data of the system. For a large quantity of biological data and complex characteristics, the paper designs a separation mechanism to separate the loading logic from the system main logical based on AOP theory. And it shows that when the framework using the mechanism, the system loading time and runtime-resource are sharply decreased by 58.9 % and 20%.Third, the implementation of the retrieval model. Its main function is to execute the retrieval task on the loaded data instances. Based on the SPARQL and OWL technology, this paper designs a semantic retrieval model for PBMICE-SR framework. Test analysis shows that the results from the model are more definite and correct.Finally, the focuses of this paper list as follow. First, this paper designs the PBMICE-SR framework with the integration technology of data extraction, data classification, data marking, data loading etc. Second, this paper put forward a mechanism for separating the edge logic from the system main logic. Last, this paper designs two models, DAGSVM-BM25 data extraction model and PBMICE-SR retrieval model.