Friday, November 06, 2009

Search v. Browse v. The Fractal Academic Library Website

In my last post, I told you that I have been trying to wrap my brain about accommodating "search-dominant" users. I'm still trying to figure out how to best serve both 'searchers' and 'browsers' for MPOW's new website in the works.

The University of Michigan has done a lot of work in rethinking and redesigning their library web presence. They have essentially distilled the University of Michigan's Libraries website into one toolbar that reads Search, Browse and Help. Search allows the user to dig right into a search or metasearch while Browse takes the user to select a subject and from that selection a dynamically generated page of different types of resources is returned including a "subject guide" to that subject area if one is available. I really think they are on to something here, although I admit that I think the user would be better served if they were taken to the librarian selected materials of the "subject guide" first.

Rather than go into the reasons why I think this, I am going to ask a rhetorical question instead: where do we want our users to start their research? Now I know that most students start their research with Wikipedia or Google and this is why I'm such a huge fan of LibX, but let's just suppose that if we had our druthers, where would we take them to start their search? And the answer is, I think librarians would want to take them a "subject page."

There are a number of libraries that try to take students down this path right from the library's homepage. The libraries of the University of Alberta list general subjects in their left margin under the heading Browse. Other institutions don't use the word 'browse' but like the University of New Brunswick allows users to find 'recommended resources' via Subject and Course Guides.

I have been working on an idea that every library "subject page" should be re-imagined as the front page of a library dedicated to that subject. The goal would be to become a page that you would imagine a student bookmarking for most of their research needs. Right now, if a physics student bookmarked MPOW's Physics Resources page, they have links to most of the things they might need from the library, but there's not a direct link to our library catalogue. U of M's library webpages are close to this vision because now, every page on their library website has a link the library catalogue among their other many resources..

But there is also a third way which hasn't completely manifested itself but I think might show some promise. At Access 2009, Bess Sadler of the University of Virginia Library, spoke about the work of Blacklight, "a faceted discovery tool." What I find most striking about Blacklight is that allows relevancy ranking to be adjusted by librarian suggestion.

When I first saw Bess' presentation, I became curious to see if this meant that a library could have different versions of Blacklight so that a particular discipline or audience could have different items weighed differently so they would get more appropriate results. So I sent an email to Bess and she kindly replied with this,

What you suggest is not only possible but one of our major use cases. We're already doing it at U of Virginia.

Here's the main catalog instance:
http://virgobeta.lib.virginia.edu/

And here is our music portal:
http://virgobeta.lib.virginia.edu/catalog?portal=music

The music portal (music view? music lens? We're still struggling with what to call these) is a view on exactly the same information that's available through our main catalog instance, but it's tailored for music scholars. The facets on the left are slightly different; for example, they contain a musical instrument facet and a composition era facet, two facets which our music users identified as crucial but which might not be especially helpful for non-music users of our collections. We also use a slight different relevancy ranking algorithm for our music portal. On our main portal, we assume that an exact match on the title is likely to be the most relevant item (i.e., we give a lot of extra relevancy weight to titles), but in the music view the first thing you type is more likely to be a performer or composer, so we more even distribute relevancy weight between title, author, performer, and composer fields.

We would eventually like to also create portals for our health sciences community, law, art, engineering, and I suspect that once these catch on we'll get many more requests.

This could be a way to bring the "recommended resources" of our "Browse" webpages to our users who only want to "Search".

No comments: