Schema Mapping Management In Peer-to-Peer Data Sharing

As the scope of social collaboration continues to expand, the need for inter-organizational data sharing application is increasingly urgent. This data sharing scenario has three main features of autonomy, dynamicity and scalability, making regular data sharing methods unable to achieve satisfactory results. Peer data management systems employ decentralized design. Schemas are set up among organizations on demand, without global schema as a mediator. On the mapping path composed of connecting schema mappings, queries are reformulated and forwarded for multiple times to achieve data sharing among multiple data sources. Although this design meets the demand of autonomy and scalability, it has inadequate response to the dynamicity. Mapping path will break due to autonomous departure of node, damaging the scale of data sharing. Three complementary improvement methods to respond to system dynamicity are presented here by managing schema mappings. With backuping schema mappings along the mapping path in advance, multiple local query reformulations on behalf can form a mechanism to bypass departure peers.Simulation and analysis results showed that, in addressing the issue of mapping path breakage caused by peer autonomy, it can achieve better results than the existing system. The number of query achievable peers increased. But this approach performs multiple reformulations on behalf, only when mapping path breakage occurs, resulting in the delayed response time.The proposed mapping composition method can deal with schema mappings expressed in a subset of XSLT correctly. After the produced direct mappings are added into the mapping network, it can not only make the network topology robust, so that the upstream node sharing of data is no longer impacted by departure peers, but also make response speed to queries greatly accelerated. However, this approach is limited to dealing with a subset of the XSLT mapping and constitutes a restriction to the map designers. Some templates have too complicated patterns, making the search for a corresponding a very complicated, time-consuming process and even failures will occur.Using virtual composition expression method to XSLT mapping not based on template matching processing model makes the expression of mapping no longer restricted by the subset of XSLT. Mappings can be composed under different circumstances, by real composition or virtual composition. The integration of instance mappings makes more consistent management of mappings. Transformation filters design can rapidly handle mappings expressed by multiple steps.