Looking at my alma mater’s library website, I noticed that they are implementing a new federated search tool that searches eight resources (a mix of A&I databases and full-text). It’s powered by Metalib from Ex Libris, a company I think is pretty great. I was really excited at first, until I decided to test it out. It’s kind of an odd bunch of databases. While a few offer articles on a wide variety of subjects, most were not the ones I’d pick for a federated search (ArticleFirst, Biography Reference, ERIC, InfoTrac OneFile, Literature Resource, OmniFile FT Mega, ProQuest Global, and PsycInfo). Maybe their choices were fueled by necessity (compatibility) rather than comprehensiveness or any particular strategy.

So, to test it, I typed in an appropriate keyword “federated search”. The federated search tool took me to a page with a list of databases and the number of results found in each. Well, actually, it took me to a list of databases and the results came up over the next minute. The results ranged from zero to 36 to over 39,000. 39,000? I looked at those result and found that none of the results on the first page were actually about federated search at all. In that database, you obviously need to do a more advanced or narrowed search to get targeted results. In fact, all but two of them offered results where the majority had nothing to do with federated search. It seems that some put the results that matched the title first while some didn’t have the search terms in the title at all. Consistently (I also tried “blog”, “dostinex”, and “pacificm”), there were two or three databases that offered me useful results and the rest were pretty useless. To their credit, they do allow you to refine the search, but I don’t know how many college students are really going to do that.

Almost every time I try to search for something at least one of the databases was not searched. Why is that? I can understand the zero result, but sometimes I sat there for several minutes waiting, and the results never came up. That’s another thing: if you think you can put some results up in a few seconds and expect college students to stare at that page for over a minute (while the other results come up) without clicking on a result, you’ve got to be kidding. People will not wait for the slower databases to show results, and that makes those slow ones rather useless. Also, I know it’s not easy to combine results from diverse databases, but it’s annoying to have eight different databases to click on to see that only a small minority of them have any useful results.

When I found something about blogs that I was interested in (it was in Wilson’s Full-Text Omnifile), I clicked on the link that said “OmniFile FT Mega (Wilson)”. I’d expected to be taken to the full-text – being that this is a full-text database – but I ended up in Wilson’s search interface! Only by going through SFX, which is another 2-3 steps, can you actually get to the full-text. Ridiculous!

I really think for federated search to be useful in any way, you either need to offer a bunch of very general databases (Wilson Omnifile, Expanded Academic ASAP, etc.) or you need to collect databases in one subject (science, business, social science, humanities). When I tried to do a search on a medication, only three of the databases gave me results (and only 6 each). On the other hand, if you’re searching for something historical (like the Vietnam War), you will end up with too many results. And the searches don’t offer targeted results where the first page of each search would always be full of useful material (rather than irrelevant retrievals). It doesn’t work for those who want advanced search features and it doesn’t work for those who are just looking for a few articles for a paper. It’s a one-size-fits-none approach. And I’m not saying that it necessarily has to be that way. I think federated search could have a lot to offer if the people developing the technology can work out the kinks. I’ve always been a fan of the idea, but the execution has left a lot to be desired.

I think this federated search tool is only going to make students more confused and annoyed with library research. Why use that when they can use Google? I think my alma mater would be smarter to push SFX more heavily and to explain to students how it works. Even I didn’t know what SFX was for the longest time because there wasn’t an explanation at each citation. How about a “full-text here” button rather than an “SFX” button? With SFX, students can find one good article in any database and easily get to most of the articles it cited. Or they can go to an I&A database and get the full-text of nearly all the results they find. It’s just so much easier. What I’d like to see when you click on the “SFX” button, though, is the full-text or an ILL page if the library doesn’t have it. It is absolutely annoying to go to that intermediate SFX page where you can choose which database provider you want to get the full-text from or you see that the library doesn’t have it. No one cares which provider they get it from as long as it’s full-text. Would it really be such a problem for SFX to just take the user to a full-text instance of the article???

I think my alma mater would be a lot smarter to just add Google Scholar to their research databases and use SFX to allow students to click through to the full-text of each article. Lots of schools are doing it. There’s even a Firefox Extension for adding the OpenURL buttons to Google Scholar results! Very cool stuff! For now, that’s a far better option for finding materials from diverse databases than MetaLib. Plus, it’s free!

Sorry, I know I’ve been rambling and ranting, but I really think libraries could be providing better access to their online and print materials. Creating easy access to materials is just as crucial as acquiring good materials, because your collection is as good as useless if your patrons can’t find it. And we need to get over this idea that it’s the patron’s fault for not being able to find things, and that all they need is education. Why should the student have to conform to an imperfect system? Aren’t we all about user-centered design in libraries? I am all for information literacy instruction, but I don’t think we should have to teach our users how to do a basic search and I don’t think they should have to know how the information retrieval system works. The workings of IR systems should be invisible. They should be easy to use. They should be designed based on how our users search the Web. I think libraries have a lot to learn from Google and from how our users use Google. Libraries shouldn’t “go Google” but they could combine the highly structured metadata of the library catalog with the highly usable search capabilities of Google. No more LCSH subject searching. No more left-anchored searches. Search engines can ignore articles rather than giving zero results when you search for “The Brothers Karamazov”. Many libraries have gone beyond these silly limitations, but many — even large University libraries — are still stuck in 1992. Look at RedLightGreen as an example of a union catalog that works. Since the vast majority of people have trouble putting their information needs into the “right” words, it should be all about information discovery. And RedLightGreen is the first step in the right direction.

If you’d like to read a much more systematic and rational post on this subject, check out Lorcan Dempsey’s The Integrated Library System that isn’t.