Keyword-based Structured Data Query Method In Dataspace

At present, we have been witnesses to a diversified and high-speed increase of information in Dataspace, people focus on the way of obtaining information, not the source of information. Because of massive, heterogeneous and distributed characteristics of these data,it becomes a serious challenge for us to how to obtain useful information quickly and efficiently in Dataspace. Convenience of keyword search make it the primary way to access these data. Therefore, it is necessary for us to study the keyword-based query methods.Keyword search is popular with users for its simply and easy-to-use and has been successfully used in relation databases. The query method liberates users from technical details, since it does not require users to grasp the Structured Query Language and underatand the underlying structure of database. There are several techniques to process keyword search over relation databases such as Steiner tree, Tuple units and Candidate Networks. However,they return tuples or tuple trees which only contain the query keywords. So they fail to provide comprehensive information about the query keyword. Sometimes users may be more interested in the information about the given keywords, the query results are no longer confined to the single tuples,and they may be some tuple trees consisting of joining tuples from multiple tables, in which some tuples may not contain any query keyword. Therefore,the query result returned by these methods can't meet the demand of the user query.To deal with user needs mentioned above, we propose a Instance Summary-based Query Answer Method(ISQAM),which will return a Instance Summary Tree(IST),including some potentially meaningful information. ISQAM also introduces the concept of affinity for filtering out relation and attribute to produce ISTs,in this way it reduces unnecessary costs. In contrast to conventional keyword search, IST contains all the information about the given keywords and enriches results. In addition, the paper also presents a comprehensive sort function to sort ISTs, then returns these correlational ISTs with respect to query as answers.Finally, we conduct a chain of expeiments on both large and small data sets and the results show that IST returned by ISQAM achieves a higher quality.