Rethinking Universal Searching
I was never a fan of universal search engines, after all I am a librarian and according to my husband anyway, I think differently. I find using MeSH and other controlled vocabulary systems to be extremely powerful when searching medical databases. I get frustrated and annoyed by various medical database vendors' universal searching, natural language searching, or basic keyword searching. After all, the medical databases I search were built upon controlled vocabulary and hierarchical subject headings and trees. I am fluent in MeSH, CINAHL, and EMTREE. Indeed searching these databases using the controlled vocabulary is almost like using another language. And there in lies the problem.
Regular searchers don't speak MeSH. For example: Until recently (2008) "cerebrovascular accident" was MeSH speak for "stroke." The term "pediatrics" refers to the medical specialty concerned with maintaining health and providing medical care to children from birth to adolescence, NOT the age group. These are just a few of the many examples of how the MeSH controlled vocabulary is not something regular library users would know.
Many medical databases vendors have built their systems to adjust for this. Many use some form of automatic term mapping to either force the searcher to choose a MeSH term or in the case of PubMed it automatically chooses terms and searches keywords unbeknownst to the user. Sometimes this process works very well for the user, other times the user scratches their head trying to figure out why "the stupid thing keeps spitting out more terms on a new screen."
Gone are the days when the librarian was the medical database and information gatekeeper. Doctors, nurses, and patients are hopping on the Internet searching for answers. Usually they come to the library when they want something beyond what they find on the Internet. They may want results that are beyond normal Internet results, yet they want the search process to be similar to Internet search or Google searching. Many of our medical databases aren't naturally set up for that kind of searching, that is why many of the vendors started adding universal searching, natural language searching, or basic keyword searching. Not surprisingly this created two different types of searchers that database providers must try to please, the expert searcher (medical librarian), and the basic searcher (normal people).
As a medical librarian I would tell you that I never use universal searching, natural language searching or basic keyword searching. Of course that would be a big, fat, red lie. I love Google almost as much as I love MEDLINE (remember my husband did say I think differently than normal people) and I use both almost equally. I use Google to search for things that are particular stumpers in MEDLINE. Often it starts my brain thinking in a different direction and I apply that to a new MEDLINE search which might yield different or better results.
I also use Google to search for non medical resources, and it is during these times that I am most like the average searcher. I am searching something very similar to a universal search engine to find information on subjects which I am not an expert searcher. Sometimes I am looking for an odd news article and while another journalism, public, or academic librarian may know the exact news database that indexes that specific paper, I don't and I don't necessarily have access to it. So I use Google. Other typical Google searches could be on home repair, business information, investments, minor questions about the law, etc. It is at these times that I can understand the easy searching appeal universal searching has for normal library users searching for medical information.
As I mentioned, more and more database producers are creating "simpler" search features for the average person. Libraries are also adding universal searching on their web pages or on their catalogs. For example: OhioLINK's front page features a universal search to find "a few good articles and books." Library users can search the "Next Generation Melvyl" at UCLA Library, a universal search engine that finds articles, books and many more resources. The University of Iowa Libraries' "Smart Search" will not only find books and article but it will also find images in their digital collections including the Hardin MD Gallery Collections.
Christopher Cox said, "boundaries are being blurred between the academic and commercial Web, between library resources, between the citation and the item itself. Students have no patience with these arbitrary boundaries; they want information, and they want it now, wherever it may be located." (An Analysis of the Impact of Federated Search Products on Library Instruction Using the ACRL Standards, portal: Libraries and the Academy. 6(3), July 2006, pp. 253-267.) Libraries users want and use universal searching, librarians need to adapt to provide other methods to help users to search for information rather than freaking out that their beloved database product now has a universal search, natural language search, or basic keyword search feature. Perhaps our education efforts should slowly be changing from always teaching the proper MeSH search strategies to how to find good information fast and to know when you to ask for help when you need more.
I tease my husband that he took the dead language Latin in high school, however what will MeSH and other controlled vocabularies be years from now? Only learned expert searchers seem to be the ones still fluent in MeSH speak, the average person is not. Latin didn't really die, it just evolved into other languages. Our language for searching is evolving too.
Labels: Databases, Medline Database, Technology

3 Comments:
I enjoyed your analogy of controlled vocabulary search to learning a foreign language so much that I wrote a blog article adding to the conversation:
http://federatedsearchblog.com/2009/01/26/the-krafty-librarian-rethinks-universal-searching/
Interesting take on Latin and Search as you develop your thoughts here.
I have a few friends who I exchange thoughts with in Latin. Something about it captures what we're trying to express --much like trying to capture the essence of something in a search box.
~Lee~
TTW Contributor.
Re MESH as a possible dead language. I don't really think that is going to happen. For those of us fluent in MESH we can find things that would get lost by keyword searching.
One of my favorite "unknown" MESH terms is "time factors". I've used it to find articles on "delay of treatment" or time from stroke to TPA or anything that has a time aspect to it. Searching "delay" doesn't pull up the articles "time factors" does.
The precision that some experienced searchers bring to a MESH search is unmatched by keyword searching. What can take a hour of messing around with keywords or a Google search can sometimes be pulled up in a minute with a focused MESH search and pull up more relevant articles. Yes, universal searching is great for just a easy and quick search but for tough and comprehensive searches MESH is the way to go.
Post a Comment
<< Home