Figure 2 gives an example from our dry run of MEDLINE semantic indexing with the early demonstration client of the analysis environment. Community repositories were generated by partitioning MEDLINE across MeSH terms. Subject domains "Colorectal Neoplasms, Hereditary Nonpolyposis" and "Genes, Regulator" are chosen and their concept spaces displayed in the middle and the right respectively. "Hereditary cancer" is entered as a search term in the first concept space and all concepts which are lexical permutations returned. Indented levels in the display indicate navigating a co-occurrence list. Navigating in the concept space moves from "hereditary nonpolyposis colorectal cancer" to the related "mismatch repair genes".

 

The user then tries to search for this desired term in another community repository, "Genes, Regulator". A straight text search at top right returns no hits. So Vocabulary Switching is invoked to concept switch from one community to another across their respective concept spaces. The concept switch takes the term "mismatch repair genes" and all related terms from its indented co-occurrence list in the source concept space for "Colorectal Neoplasms" (including many not shown) and intersects this set into the target concept space for "Genes, Regulator".

 

After syntactic transformations, the concept switch produces the list (at middle right) of concepts computed semantically equivalent to "mismatch repair genes" within "Genes, Regulator". Navigating the concept space down to the object (document) level locates the article displayed in the middle bottom. This discusses a leukaemia inhibitory factor which is related to colon cancer. Note this article was located without doing a search, by concept switching across community repositories starting with the broad term "hereditary cancer".