Research On Interface Integration And Query Translation In Deep Web Based On Domain Ontology

With the massive and increasing growth of network information, more and moreWeb information is no longer stored in the form of static web page. The Webinformation is stored into Web database and hidden behind the query interface, whichis so-called Deep Web. Deep Web contains a lot of high quality information.However, because the Deep Web has the character of large-scale and heterogeneous,if the users want to obtain the content from multiple Web database, they have tosubmit their inquiries many times. This method greatly wastes users’ time and energy.So it is necessary to provide a unified approach to visit the Web databases of the samedomain. Also because Deep Web has many characters such as heterogeneity, dynamicand autonomy, which makes the equivalent query transfer between the interfaceintegration and the Web database interfaces within one domain hardly realized. Thus,the integrated of Deep Web interfaces and query transfer in special domain havebecome the hot spot of research recently.To solve the disadvantages such as low definition of integrated attributes matchand low accuracy of transfer which occur in the existing methods of interfaceintegration and query transfer, the methods of Deep Web interface integration andquery transfer based on domain ontology are studied in this paper. Firstly, theframework of Deep Web interface integration and query transfer based on domainontology is given. Then, the integrated method is proposed by the direction of theframework. The method includes algorithm of mark domain ontology, algorithm ofconcept abstract from domain ontology and the algorithm set optimization. Next, themethod of query transfer is put forward. The method contains attributes match,predicate template match and query rewrite. Finally, an experiment is used to verifythe method proposed in this paper is feasible. The main tasks of this paper are asfollowed:(1) Give the framework of Deep Web interface integration and query transferbased on domain ontology. On this basis, the functions of the core modules and theworking mechanism of the framework are introduced. (2) Put forward the interface integration methods. First of all, based on thedomain ontology, group the attributes which are need to be integrated (commonattributes group and special attributes group), and mark the attributes in commongroup as the domain ontology. Secondly, conduct concept extraction from the domainontology tree to complete the integration of common attribute. Next, optimize theintegration to complete the integration of the special attributes. Finally, obtain theattributes set of the interface integration by merging the integration results from thetwo parts.(3) Propose the method of query transfer. Above all, get match table accordingto the method of interface integration to realize attributes match. Next, design thepredicate template of attributes with constraints. In this paper, different processors arerespectively designed for different type of template. For every type of processor,while handle the predicate template matching, the process of constraints in predicatetemplate are mainly studied in this paper.(4) Design and realize the prototype system of interface integration and querytransfer. Verify the method is feasible and effective by the comparing and analyzingthe experimental results.