Research On Application Of Metadata Technology In Science And Technology Resources Database Query

This dissertation is based on the project of science and technology resources database network construction of Hebei Province, and it focuses on the problem of efficient information query among the distributed and heterogeneous science and technology resources databases of Hebei Province. A unity retrieval system which has high reliability, good compatibility and can adapt to the Internet deployment application environment is set up, and it can provide the query and browse service of the uniform access for the multiple subjects'information.According to the characteristics of the science and technology resources database of Hebei Province, a reasonable multi-layer metadata standards structure for the core database and other special databases is defined. There are two layers. Core metadata serves for the data sets. A set of elements is defined for the data sets of the whole science and technology resources database of Hebei Province. Professional metadata serves for professional data, which provides the reasonable access points for professional data retrieval. In addition, core metadata is the natural blood tie among the professional metadata and the common basis. Therefore, good interoperability is possessed.The data sets are described by core metadata standard in the multi-layer metadata standards structure so as to facilitate the navigation of resources. Data sets are retrieved by metadata. This way can narrow the search scope and improve the efficiency of view. In the data sets retrieval algorithm, three important existing tree matching models are used and the matching results of data sets are presented according to the matching costs in a descending order. The function of matching pretreatment is added to improve the existing algorithm, and a series of experiments are conducted. The results show that this algorithm can remove the unwanted node in the tree, and promote the efficiency of data sets retrieval when the data scale is very large. Especially, the recall rate, the rates of accurate survey and the average response time are satisfying. It is necessary for the sorting method of retrieval results to be improved and optimized in order that the users can find the resources they need as soon as possible. Vector space model is constructed between the query conditions and the query results. The results of the single field retrieval are presented in a descending order according to the similarity degree between the query conditions vector and the query results vector. In order to make up for the traditional vector space model inadequate, the function of query, expansion is added, and it can efficiently solve the problems of low recall rate which result from the synonyms mismatching. In addition, the calculation of auxiliary weights is utilized to refine the search when the similarity degree values of multiple vectors are the same.At last, the dissertation is summed up, and the direction for further research is pointed out.