Federated Searching Idiosyncrasies

Last night I taught a BI session, and after demonstrating a couple of databases, a bright and impatient student asked if we had a way to search all of the databases simultaneously with the same search. We do in fact have a federated search function, which after three years is still labeled a pilot service on our website. (I would give you the link, but you wouldn’t even be able to see the page without being on the campus network.) I rarely mention the federated search engine because I haven’t found it very useful. I said that theoretically we did have a way to do this, but that I didn’t like to use it. However, let’s try a search and see what happens.

We were searching theater (chin* AND theat* AND brecht as keywords was the search if you want to try this at home, boys and girls) and had found 313 records in the International Index to the Performing Arts using this search before narrowing the search some. The federated search on the theater database page simultaneously searches the IIPA, American Drama, MLA, JSTOR, Oxford Reference Online: Performing Arts, the Essay and General Literature Index, and the World Shakespeare Bibliography Online. The student was running the demonstration computer, so he did the search. Despite all the databases being selected, it searched only American Drama, and found nothing, naturally.

Then several other students did the same search, and each found a different result. The other students did all manage to get results from the IIPA. One got 64 results, another 101, another 152, and another 313, which is the number we got from the direct search. Needless to say, it was weird, but made my point that while there are some benefits to federated searching, there were also some problems. It’s possible not everyone was doing the same search, though they claimed to be.

It’s really too bad, because I was a proponent of the federated search feature before we implemented it, and it was only after using it more that I found it less helpful than I thought it would be. I wanted it because of the simplicity I thought it would bring to novice researchers. I had in mind just the sort of student who asked the question last night. If we take last night’s search, IIPA was an excellent place to begin, as I had already told them. The federated search (when it worked at all) was good for directing them to a database that was both searchable and appropriate. Even with the varying searches, IIPA always had more than all of them except JSTOR. But to narrow the search in IIPA, one still has to go into the IIPA itself. Even trying to view the record to see what descriptors might help narrow the search takes you out of the federated search engine and into the database. At this point, instead of this search engine, I’d just as soon have Google Scholar searching through all of our databases, that is, if Google Scholar worked better.