Libr 528  Literature Searching

 

Gold Standard Searches

 

            Recently the NLM Technical Bulletin has issued a series of challenge searches to the medical library community.  These challenge searches are "real" searches that originate in the day-to-day experiences of medical librarians.  The search question is offered as a challenge to the medical library community, and searchers compete for the "best" search.  NLM searchers also do the searches and act as arbiters of which are the "best" searches.  "Best"  would be defined as the search that captures the greatest number of prima facie relevant documents, where the arbitrary size limit on the retrieval set is 50.

            Often two or more searches are determined the best, or equally good.  And their search strategies may be strikingly different.

            Following are the search strategies for these gold standard searches.  Note that the query language used is Orbit:

 

            1. The role of the National Library of Medicine in the development of user-friendly computerized information retrieval systems.  NLM Technical Bulletin, July-August 1991, (261), 24-31.

 

            2. Is the risk of miscarriage increased by working at a video display terminal? NLM Technical Bulletin, November-December 1992, (269), 6-13.

 

            3. Predictors of long-term success in the maintenance of weight loss. NLM Technical Bulletin, July-August 1993, (273), 20-25.

 

            Your challenge: You are challenged to a better search that the Gold Standard Search.  Pretend that you are a reference librarian and your colleague shows you the Gold Standard Search in the NLM Technical Bulletin, and you scoff, "I can do a better job than that!".  Now, prove it.

            Limitations: (1) You must search in Dialog, and (2) You must begin your search strategy in File 411,  Dialindex.  You may use any Dialog database that you can justify as appropriate.

            Hand in:  Justify how and why your search is "better" than the Gold Standard Search.  Higher quality (i.e., "better") might be some combination of database selection, term selection, syntax elements, searching strategy, and finally that your percentage of prima facie relevant documents is as good, hopefully better, than the Gold Standard Searches.  Do not hand in any of the documents.

            You may have to make some judgments about interpretation of the questions, but that might be simply part of "proving" why your search is better.

            Starting Point:  Each proof of higher quality must begin with your manipulation of the elements of Dialindex.

            Grading:  Your efforts will be evaluated on the basis of how well you understand the strategy used by the Gold Standard Search and how your search is demonstrably better. 

            I suggest you pause for a moment to consider how you would build an argument that your search is "better" or at least "as good as" the Gold Standard.  Elements of such an argument might include (1) Your databases are more appropriate for the given question being searched, (2) Your search benefits from the actions of Dialindex, (3) Your search benefits from the actions of OneSearch, and (4) Your search benefits from the actions of Dialog's Target, Rank and Duplicate Detection Commands.  There are also actions you might take using Dialog limiting commands to one language, say, English.  You might also consider Dialog's Current command to make sure that your set of approximately 50 records are the only the most recent.  I would suspect that these features would be the minimum required parts of a substantial and credible argument that your  search is better than the Gold Standard.  There well could be other features and arguments that you might want to consider as well.

 

Some strategy notes [suggestive, not definitive] on making the argument that your search is “better” than a Gold Standard search: