DeepSearcher: A One-Time Searcher for Deep Web

Derong Shen, Gaoshang Sun, Tiezheng Nie, Yue Kou

Abstract: The proliferation of database-driven Web sites has made user pay more effort for selecting the best satisfying results. Therefore, we propose a searching system named as DeepSearcher to meet userpsilas need, which includes offline processing (e.g. pre-processing) and online processing. The latter consists of query processor, result integrator, cache subsystem and service portal. To implement the system, key techniques such as subject-based classification, clustering-based result extraction and schema recognition, dominant attribute-based data sources ranking, query relaxation, duplicate identification and result top-k are adopted to support the searching system. The demonstration shows the feasibility and the promise of DeepSearcher.