Answering approximate queries over autonomous web databases

Improving web database search incorporating users query information

“…In both cases query reformulation is a normal behaviour of Web users [5]. To provide relevant information to users various methods like query reformulation [10] and query relaxation [8], [9] have been proposed, however, these methods have not lessened the burden of searching relevant information as needed by a user. Query reformulation may often distract from actual search needs, while query relaxation may again return too many answers.…”

“…In another ranking method [14], a database keyword based search is proposed by indexing the related information about tuples. In other work [9], methods of tackling empty answers have been proposed by rewriting the users query and relaxing constraints.…”

“…In order to return relevant results for users query, many methods are used like collaborative filtering techniques (which group similar users/query), [10], query refinement [9] and utilizing user profile [1], [10]. However, universal methods that cater to multiple problems and utilize all available information are still lacking.…”

Combining Approximation and Relaxation in Semantic Web Path Queries

“…[2,17], as have approximate query matching techniques, e.g. [4,7,14,15,18]. In [13], we studied a combination of these and showed that approximate matching of CRP queries can be undertaken in polynomial time.…”

“…More generally, [7] used weighted regular transducers for performing transformations to regular path queries (but not CRP queries) to allow them to match semi-structured data approximately. The approximate queries of [18] are simply selections placed on attributes of form-based web data, where value constraints can be relaxed according to their perceived importance to the user. In terms of query relaxation, work has been done on relaxing tree pattern queries for XML, recently in [16].…”