[sigia-l] Faceted browsing WAS The category of "Miscellaneous"

[Avi Rappoport]

There are a number of alternatives to Endeca.  The two I know most 
about are i411, which scales to the millions and is running a baby 
bell yellow pages site,<http://www.dexonline.com>, and Siderean 
Seamark, which uses RDF (links at 
<http://searchtools.com/info/faceted-metadata.html>).

I think the point of faceted metadata providing rich access to 
"miscellaneous" data is a very important one.  Facets such as author, 
institution, brand, or format are ways to get around the limits of 
the traditional subject hierarchy model.  In my view, Miscellaneous 
will always be around, because there will always be information 
that's not part of the core of a site but is related in some way. 
Faceted metadata search can incorporate these otherwise isolated 
pieces of information.

Avi


At 9:33 AM -0500 6/16/04, Jonathan Broad wrote:
>First, let me say that I'd be very interested in any information 
>about alternatives to Endeca people might know about.  So far as 
>I've been able to tell, it is just about the only provider of 
>enterprise-scale faceted browsing.  Flamenco's open source project 
>probably won't scale to the data set I'm currently working on 
>(~500,000 items, with a ~10-15,000 term controlled vocabulary).
>
>On Jun 16, 2004, at 4:54 AM, Donna Maurer wrote:
>
>>Having spent my day up to my elbows in faceted browsing (which also
>>involved creating a new page on the IAwiki:
>>http://iawiki.net/FacetedBrowsing), I wonder why you think that it
>>eliminates the dead ends.
>>
>>All of the current implementations of faceted browsing I have found
>>purposely lead to dead ends. Faceted browsing is a filter, and a filter
>>is a funnel that progressively narrows the number of results down.
>>You browse through the facets until you find a product/page that is
>>interesting. Although it would be simple to use the facets to create
>>relationships (and eliminate dead ends by expanding the results), I
>>haven't seen this done at all.
>
>Sorry for the mix-up:  by "dead end" I meant something similar to 
>these aspects of FB noted on your wiki page:
>  * users filter a set of items by progressively selecting from *only 
>valid values*
>*  it is impossible to get a *null result*
>Endeca makes this elimination of "no results" (perhaps a better way 
>of putting it) a central feature of their marketing.
>
>But I think that both Endeca and the Flamenco interface allow you to 
>shift strategies without needing to back up a hierarchical branch.  
>Even the final result-page in the Flameno interface allows you to 
>remove any currently applied facet, start a new search from any of 
>the result's facet-values, or start a new search based on the 
>ancestors of any applicable facets.
>
>The basic idea of faceted browsing is the application of reductive 
>filters, but all the implementations I've seen benefit from the 
>ability to add or remove filters in any sequence, and at any stage 
>in the search process.  Ideal for berry-picking.
>
>Jonathan

-- 
   Avi Rappoport, Search Engine Consultant <mailto:avirr at searchtools.com> 
   Complete Guide to Search Engines for Web Sites and Intranets
             <http://www.searchtools.com>