Interesting site, code4lib. I followed a link to a presentation given by Erik Hatcher for the Library of Congress on August 1, 2007 about Solr, a search engine for libraries, based on open-source Lucene.
In it he discusses the development of a new search engine utilizing the Java based Lucene open source software with Solr (originally developed by CNET, and open-sourced) layered above. The resultant product has caching, replication, faceting, hit highlighting, spell-checking and an htttp interface. It is being used at the Smithsonian, Internet Archive, Open Library, UVA (where Hatcher works) and Peel's Prairie Provinces.