With Greg and Martin discussed with Karin possible framework technologies to support this project, which is essentially a meta-search engine which does a lot of indexing by calling each collection using whatever API is made available and the providing the user with a query interface which returns results from the index and links to the original source and its search interface where possible.

PHP, Java likely candidates for coding.
Postgres or Oracle backend, considering scale.
Also talked about XUL as front-end, especially the idea of assembling a custom search-engine interface using a wizard