Information Extraction from Template-Generated Hidden Web Documents

The larger amount of information on the Web is stored in document databases and is not indexed by general-purpose search engines (such as Google and Yahoo). Databases dynamically generate a list of documents in response to a user query – which are referred to as Hidden Web databases. Such documents are typically presented to users as templategenerated Web pages. This paper presents a new approach that identifies Web page templates in order to extract queryrelated information from documents. We propose two forms of representation to analyse the content of a document – Text with Immediate Adjacent Tag Segments (TIATS) and Text with Neighbouring Adjacent Tag Segments (TNATS). Our techniques exploit tag structures that surround the textual contents of documents in order to detect Web page templates thereby extracting query-related information. Experimental results demonstrate that TNATS detects Web page templates most effectively and extracts information with high recall and precision.