Re: [TaxoCoP] A question for everybody. How easy is it to create a faceted Taxonomy?
- Adding on to the excellent advice, a couple of points to be aware off:
1) Understanding excatly what is the purpose of the facets and how
they are going to help the end users AND ALSO how it benefits the
business. It's a simple point , but 3/4 of the implementations I've
been involved in ( they always call you in only in the last minute)
the business or the technical team can't seem to articulate the
purpose of why they are using facets at all....
2) Understanding that there are two parts to developing a faceted
browsing / searching system. There is the interface aspect.......which
most people don't seem to be aware offf...and then there is the facet
development aspect. If either of these are not taken into
consideration , your project ALWAYS ends up a failture.
3) Updating the facets should be part of the project on how maintenace
should be done.
4) Lastly, usability testing should always be part of the project from
the scoping to the maintenace phase.
Very high level points, but things I felt needed mentioning.
On Wed, Oct 1, 2008 at 7:30 AM, Avi Rappoport <analyst@...> wrote:
> Not only should it be it easy to add facets, but it's a good idea
> allow additional metadata to spawn facets (or at least sub-facets)
> dynamically. That is, if you have a facet for price, and you have
> ranges from $1 to $200, adding a $300 item should just update nicely,
> without much, if any, work. If it requires significant effort, your
> facet-handling tools may be misconfigured or inappropriate.
> Trying to anticipate all possible facets can be a huge time-sink as
> well. I've seen so many installations where the project managers,
> domain experts, and even taxonomists have gone ahead with their
> expert opinions instead of starting from user testing, and iterating
> several times based on actual use. If the experts guess wrong, a lot
> of effort is wasted. So I'd say start simple.
> At 10:16 PM +0000 9/30/08, alexkatsanos2 wrote:
>>Specifically, if we create a faceted layout, then how difficult would
>>it be to add more facets later?
>>That way we wouldn't have to go so crazy trying to make sure we have
>>every single facet imaginable whether or not we need them.
> Enterprise Search Analysis -- Search Tools Consulting
> (510) 845-2551 / analyst8@...
> Complete Guide to Search Engines for Web Sites and Intranets:
- At 10:15 03/10/2008 -0400, Eric Erickson wrote:
>Library items are only classified once, but digital resources andMay I de-lurk for a moment, please, to try to clarify something? My
>items in the catalogue may have multiple tags.
comments aren't directed at you specifically, Eric - yours isn't the
only post (on this and other lists) to give me the impression that
the functions of traditional library classifications aren't well
The point that so often seems to be missed is that library
classifications were designed to do two things at once - identify the
subject matter contained within a physical item, and identify the
location of each item on the shelves. Taxonomies for electronic
resources are concerned only with the first function - information
retrieval. To compare them to library methods of *item* retrieval is
inevitably going to cause confusion.
"Library items are only classified once" is true if:
a) the classification is hierarchical (to librarians that means ones
like Dewey and LOC);
b) the classification is viewed purely as a device for item
It's true, in those circumstances, because of the obvious fact that
each *physical* item can have only one location in a classified
In the case of faceted classifications, the statement is partly true.
These classifications do allow you to 'tag' an item with multiple
descriptors, but in non-digital libraries the codes for these 'tags'
are then combined in a specified order to produce a single code
identifying the location of the physical item on the shelves.
In both cases, the statement is not true for the library function of
information retrieval. The *content* of a library item can, as Paula
says, be given multiple access points within the catalogue, and these
access points may be sorted in any order thought useful. True, those
'tags' are attached to a surrogate (the catalogue entry) rather than
directly to the physical item itself, but that doesn't stop them from
fulfilling a similar role in content retrieval.