Research On Keyword-Driven Construction Of Structured Queries

With the advance of the Semantic Web, inceasing amounts of data are available in a structured and machine understandable form. This opens opportunities for users to conduct semantic search based on it. However, users have to pay a high price for constructing structured queries and prefer keyword search, keyword search on structured data has become a hot research topic. But because of the inherent ambiguity of keyword search, the challenge of adapting keyword search to semantic search is huge.Based on previous research, this paper has studied the interaction and search of the keyword search on structured data, in particular RDF data.Ini this paper, the way of converting keyword queries into structured queries is used, instead of computing answers directly from keyword queries. After getting structured queries, the users are to choose the appropriate query which is processed by the underlying database engine to get the answers. The main contribution of the paper contains two aspects:new sorting algorithms for candidate queries, rational and efficient interaction with users. Based on it, this paper proposes two ways to adapt keyword query to structured query on interaction and queries sorting. Compared to current solutions to Star Query(a limited structured query), method one predicts certain categories of each keyword from user input, gets ranked combinations of keyword with some category based on a new query ranking algorithm, and finally return the corresponding information needs through user interaction. Method two, which applies to a universal structured queries, alleviates the bad impact from the keyword query inherent disambiguity on adapting keyword queries to sturctured queries by conducting interaction with users on keyword mapping, and achieve promising performance improvements compared to other approaches based on taking the user’s keywords order into account.Finally, two prototype systems have been implemented in light of above-mentioned two methods. The experiment evaluation validates that they can get respectable result in both effectiveness and efficiency.