Intelligent Proxy

• "Pass through" proxy that rewrites the HTML page before returning it to user
• A&I component knows how to parse the page returned from A&I search, insert a "Search for document" link (see before and after screenshots of PubMed). Different A&I databases can require different components. A&I component passes along citation information in internally consistent format (tagged text for now).
• Proxy uses citation information to query local "holdings database" to find fulltext provider for the desired issue.
• "Fetch fulltext" component knows how to fetch fulltext from a particular fulltext supplier, given citation information. Different fulltext suppliers require different components.
• Integration with other services allow us to provide appropriate fallbacks (Article Delivery).
• Prototyped using PubMed, Academic Press/Ideal, and Springer.

Holdings Database (Digital Registry)

• Searching for fulltext is slow, so we want to know where (and if) to search for it before trying.
• Need to maintain detailed holdings data for each title (MARC Level III holdings) to support that decision-making.
• Database provides place to keep URLs for components associated with each A&I and fulltext provider
• Database provides place to keep userids and passwords associated with each A&I and fulltext provider (components can query database to obtain them).