This seems obvious, but I'm surprised how infrequently it's actually used: I think the key to designing good advanced search screens is not to be tied literally to the underlying data structure. The boxes/options you show on an advanced search screen don't have to be a 1:1 correspondence with fields or elements in the data. Certainly you don't have to use actual field/element names as labels in the interface - show user-friendly ones. But more importantly you can group elements together. Have a defined use case for every field you put on that screen, and have it search one or more data elements that meet that need. It's perfectly OK to have a box on that screen search more than one field on the back end. In the case of MODS, that's been cited in one email in this thread, combining the different date elements within originInfo into a single index likely makes sense. You might also make them searchable individually, but only if the target users are likely to need that (very granular) distinction. You could also build a combined index on all names, or in addition provide indexes for names only recorded as having a certain role.
And don't be afraid to leave fields out either - only put fields on the advanced search screen that make sense, that have a reasonable (but imaginative!) reason for being there. Using MODS again, langugageOfCataloging would likely be something you'd leave out of all but the most exceptional advanced search screens. But there are probably many more borderline cases that deserve careful consideration. It's hard to find a good balance between throwing in the kitchen sink and meeting advanced users' needs.
One other piece of advice: use lists of defined values whenever possible, but be careful about only making some things (language for example) available as search limits rather than as searches in their own right - why not allow someone see every resource in a certain language? If they get more hits than they're comfortable with, they'll just use your (well-designed, of course) feature to revise their original search.
Jenn
========================
Jenn Riley
Metadata Librarian
Digital Library Program
Indiana University - Bloomington
Wells Library W501
(812) 856-5759
www.dlib.indiana.edu
Inquiring Librarian blog: www.inquiringlibrarian.blogspot.com
> -----Original Message-----
> From: Code for Libraries [mailto:[log in to unmask]] On Behalf Of
> Mark Jordan
> Sent: Friday, November 14, 2008 5:01 PM
> To: [log in to unmask]
> Subject: Re: [CODE4LIB] Good advanced search screens
>
> Hi David,
>
> You might want to consider an advanced search interface that offers a
> varying number of options. We've done this to a certain extent in the
> PKP Metadata Harvester for schemas more complex than Dublin Core. An
> example of a harvester that has some MODS in it is at
> http://harvesters.sfu.ca/chodarr/index.php/search, if you want to see
> how we implemented this (click on the "More fields" button).
>
> We're currently rewriting the Harvester so I'd be interested in hearing
> what you settle on. That particular application suffers from the same
> problem you're describing with WorldCat -- a very rich metadata set to
> search against, plus in the Harvester's case, new schemas can be added
> fairly easily, and we don't want admins to have to rewrite the search
> form when they add a new schema.
>
> Mark
>
> Mark Jordan
> Head of Library Systems
> W.A.C. Bennett Library, Simon Fraser University
> Burnaby, British Columbia, V5A 1S6, Canada
> Voice: 778.782.5753 / Fax: 778.782.3023
> [log in to unmask]
>
> ----- "David Walker" <[log in to unmask]> wrote:
>
> > I'm working on an advanced search screen as part of our WorldCat API
> > project.
> >
> > WorldCat has dozens of indexes and a ton of limiters. So many, in
> > fact, that it's rather daunting trying to design it all in a way that
> > isn't just a big dump of fields and check boxes that only a cataloger
> > could decipher.
> >
> > So I'm looking for examples of good advanced search screens (for
> > bibliographic databases or otherwise) to gain some inspiration.
> > Thanks!
> >
> > --Dave
> >
> > ==================
> > David Walker
> > Library Web Services Manager
> > California State University
> > http://xerxes.calstate.edu
|