E-government information can be difficult to find, access, and use because much of it is hidden in the deep Web, e.g. dynamic content automatically generated from databases. Although commercial and governmental search engines and directories provide access to much Federal information, significant challenges remain. Techniques for exposing Deep Web portal pages are not nearly as well defined as the field of full text searching. Since databases typically have far more dimensions than regular web pages, the question is how to create a topical index that is detailed enough to get potential users to a site portal page, but not so enormous as overwhelm users with a large set of databases for any query. The access of Deep Web portals requires extra data structures such indexes. Although automatic computational software may be able to generate preliminary index words and descriptions for the Deep Web portals, the automatically generated data structure is highly noisy and inaccurate. In this project, the team plans to solve this by combining automatic information processing with social computing mechanisms, particularly collective knowledge accumulation and collaborative sense-making ? social cognitive processes by which people collectively gather, organize, and understand information and knowledge. By introducing social computing mechanisms, the challenge of designing incentive mechanisms is introduced. These will be explored using mechanisms from incentive centered design. The major pay-off in this proposal is the improvement in transparency and information services provided by government websites, exposing resources and helping users to understand them.
