Determining Relevant Deep Web Sites by Query Context Identification

Deep web search requires a transformation between search keywords and semantically described and well-formed data structures. We approached this problem in our "In the Web of Words" (WoW) project by allowing natural language sentence queries and by a context identification method that connects the queries and deep web sites via database information. In this paper we propose a novel SQL based approach that can identify the focus of input questions if the information is represented in a database. We propose a new relational database design technique called normalized natural database (NNDB) to capture the meaning of data structures. We show that a proper NNDB is a context database, and it can serve as the basis of context identification combining the template based techniques and the world model encoded in the database. Key words: deep web search, context identification

• Type: Conference
• Year: 2008
• Where: BIS
• Authors: Zsolt Tivadar Kardkovács, Domonkos Tikk
• Keywords: BIS 2008, Business, Context Identification, Deep Web, Deep Web Search