Querying Biomedical Information on the Web

The US National Library of Medicine has more than 30 years' experience
providing world-wide online search access to large, dynamic databases of
vital, high-quality information in the health sciences.  It has developed
a very mature and sophisticated real-world controlled vocabulary, MeSH,
and has built the UMLS Metathesaurus, which integrates more than 40
biomedical vocabularies and classifications.  The NLM also conducts and
funds advanced research in informatics-intensive areas such as human genome
mapping, the Visible Human, full-text searching, and interdatabase linking.


Requirements for a Web Query Language
-------------------------------------

The NLM's extensive experience with query languages includes http servlet,
java applet, CGI, Z39.50 and telnet-style command line interfaces.  The
lessons learned generalize easily to non-biomedical information and lead
to the following (not comprehensive) wishlist for a standard web-based
query language.

  1. Should support fielded search as well as full text search.

  2. Should support boolean search with ranked results.

  3. Should identify some common search access points (e.g., Title, Subject).

  4. Should rely upon a system of open standards and protocols.

  5. Should recognize the role of controlled vocabularies and authority files.

  6. Should provide a standard way to query databases for schema information.