Sunday, 7 March 2010

Comparing Googles to oranges.

Being a librarian, I often hear comments like, "Why doesn't this (insert information searching tool here) work like Google?" I recently came up with a snappy but sarcastic retort to this: "Because Google doesn't work."

Although the Google interface (or rather the single-box keyword-searching text box that has become synonymous with Google) is very attractive in it's simplicity, users pay a price for that simplicity. For anything beyond the straightforward where's-the-website-for-that-institution-that-I-know-the-name-of, searching by keyword, even with a fancy ranking system, is fraught with dangers.

Ok, maybe not dangers, but looking at a search tool that's designed to help you perform complex and comprehensive searches through academic publications and then comparing it to a search engine to find the web page you want is like comparing apples to oranges. No it's worse than that... it's like comparing apples to the moon! Article indexes/databases like MEDLINE, EMBASE, CINAHL, etc and their interfaces (sorry, I'm a health sciences librarian so those are the resources I know) are complicated for a reason. Databases provide a wealth of somewhat standardized pieces of information that can be used easily to find the record (or article) that you want. Creating an interface that doesn't take advantage of that organization of data would be foolish.

Google on the other hand has to contend with trying to reign in the chaos that is contained in billions of web pages that don't have any real amount of standardization. The only thing that they all have in common is that they have text. That's all Google et al have to work with. And they've done a great job, don't get me wrong. But you wouldn't want a tool like that searching through journal articles primarily. Google tries to glean what a web page is about by looking at the frequency and location of words and matching them with submitted search terms. In the databases I mentioned above, experts have read the articles and "tagged" them with subject terms that describe in pretty good detail what they are about. Guessing at that point would be foolish. The problem is that the searcher then has to do a little bit of work to figure out what terms describe what they really want. But the extra work pays off in better results.

So don't wish for a more Google-like interface. Wish for a bit of extra time to do a better search.

No comments: