A Discovery Of Data Service Hyperlinks Based On User-feedback

In the era of big data, data are playing more and more important role in social and economic activities, which also have been considered as new economic asset classes. In recent years, with the rapid development of Internet technology, many government agencies and organizations (such as Data.gov, Google Table, Public Data Sets on AWS, DataOne, Freebase, etc.) have opened data to the public. It is difficult for isolated dataset to develop value, in order to achieve on-demand convergence and integration of these multiple sources of data. We need to face these data from different sources with the challenge of data structure heterogeneity, access method heterogeneity, and semantic heterogeneity.In this paper, we use data service to solve the problem of data structure heterogeneity and data access method heterogeneity. Based on the data service model, we find the data service hyperlinks by using the semantic relationship of data service’s parameter, so that we can achieve the purpose of data integration on demand by combining data service. In the existing researches, finding the semantic relationships of data service’ parameters depended on automatic matcher. But the automatic matchers by using one data attribute or a few data attributes are hard to make sure the matching result correct. Because the results of automatic matchers may be different, which is also called uncertainty. In order to solve the problem of uncertainty in the matching process, there are a number of studies based on mathematical models to fully analyze and use the different automatic matchers’ matching results to improve the accuracy of automatic matcher. However, these efforts still can’t guarantee the matching results accurate.In the big data background, this paper proposed a semantic matching method, which is interactive and based on user participation, to solve the uncertainty in the process of finding the semantic relationships of data service. This method introduced user feedback to effectively establish the semantic relationship between the input and output parameters of data services for multi-source data integration. The main work and contributions are as follows:(1) we analyze and define five semantic relationships between the input and output parameters of data service, on the basis of that putting forward data service hyperlink model.(2) We put forward an interactive semantic matching method UF-Matcher based on user feedback, and experiments show that the method can effectively improve the matching results. Experiments also show that the UF-Matcher can settle down the uncertainty and keep the balance of matching results and low user burden.(3) Based on UF-Matcher semantic matching method, we developed a discovery system of data service hyperlinks.