Federated Searching -- fad or future?

A colleague of mine, an instruction librarian in an academic library, is writing an essay on federated searching and asked my opinion of such applications. I know this is something that people at the Olin Library have been considering recently, so I thought you might be interested in my thoughts on this issue.

The University of Pittsburgh Libraries have been using a federated searching application from Webfeat for a couple of years and that is the only application with which I have any experience. Other vendors include the Open Source DBWiz, MetaLib, SirsiDynix's partner MuseGlobal, and Central Search. This is a fast moving field so I am sure there will be others and by the time you read this some of these vendors may have gobbled each other up. Welcome to the wonderful world of library systems software!

When I explain federated searching to library users I usually start by saying, think of it as Google for libraries. An easy way to search multiple databases with one search and get relevant results all in much the same format. The user can quickly and seamlessly move from search to fulltext, avoid doing the same search multiple times, using different search techniques for different databases, and then searching again for the fulltext of the articles they seek. Sounds great right? What could possibly be bad about such software?

Before I tell you my opinion you must realize that I am a librarian. That makes me very different from the vast majority of library users. I think it was Roy Tennant who said, "Librarians like to search. Everyone else likes to find." I have made a career out of understanding how to search for information and how to help others search, and I hope find, what they need. I understand how to search a variety of databases and interfaces. I even read the introductions to reference books. The bottomline is that my opinions about federated searching are probably very different from a normal library user, so take them with a pinch of salt. This is something that I find librarians forget too often.

That being said, I find that federated searching software feels like a scrim in the theater. It makes everything slightly hazy, and softens the definition of the individual information objects I seek. I am left with the feeling that I want to tear the curtain away to get to the real information in the databases below.

Information is complex, in format, content, and in the relationships between information objects. Federated searching makes each piece seem very similar to every other -- the record of a book chapter, a journal article, a video, a government document, etc. All look alike and it takes extra effort to work out what you have retrieved. The implementations of federated searching in libraries usually strive for simplicity and exacerbate this homogenization problem.

The current versions of federated searching cannot cope with the most sophisticated elements of library databases; the niceties of controlled vocabulary, or the tree structure of MeSH. Searching is reduced to relevance and keyword searching, and thus ends up being quite blunt.

I often see the results of these problems at the reference desk. Users come to the desk frustrated that they cannot find what they want after having tried a federated search, or unable to interpret the results of a federated search. I find that the most common solution for these users is to show them how to go directly to the most appropriate database for their information need, where a more directed search quickly retrieves what they want. However, only the most persistent and confident users will come to the reference desk after an unsatisfactory search. Many others will seek help from a friend or colleague, try a very different approach, or just give up. Presumably many others are perfectly happy with the results of their federated search. Librarians must be careful not to draw conclusions about the average library user based on the minority who seek assistance from a reference librarian.

I do like the ability to search multiple databases simultaneously and to quickly see which databases might reward further searching when a federated search system tells me how many hits I get for a search in each database. Selecting a database to search has always been a mysterious task for novice library users and anything that helps them with this task is to be
welcomed. I also see that my criticisms make me sound like the experienced Dialog searchers in the generation of librarians just before my own. I would listen to their war stories of finely tuned, cost-effective searches in arcane databases and their grumblings about the younger generation of wasteful searchers and secretly think that their expertise had be superseded. Perhaps, like them, I am just getting old and curmudgeonly.

One final point, I am not convinced that federated searching will not turn out to be a transitional technology, like CD-ROM jukeboxes, more an expression of a temporary limitation in the technology available in libraries than a transformational technology that fundamentally alters how people interact with information.

Still the technology bears watching and the answer for libraries probably lies in implementing such search capabilities while maintaining access to individual databases and continuing to do what we have always done, work with users individually and in groups to help them understand the many ways to search for and find information.