[Sigia-l] Metadata::Information type
Jason Burton
jason.e.burton at virgin.net
Mon May 19 18:22:04 EDT 2003
Hi,
This is my first post, so apologies if its not in keeping with the list (to
get my excuses in early - I did check the archive before posting).
I'm currently working on a site: c. 2,500 plus pages, access to a database
of offline library material, database of archived magazine articles among
other things.
We know that people are frustrated with the search facility and suspect that
part of the problem is that certain types of information - that are relevant
to some users, but not all - are swamping the relevant results.
A couple of scenarios to illustrate my point:
User 1 is a researcher looking for material on affirmative action in the
workplace - to write an article about its impact on company performance for
a thesis. She searches on 'affirmative action' and gets back some articles
from a magazine archive and some references to journal articles in the
library. She goes away happy having gathered some substantial material for
her thesis.
User 2 is working under pressure in a small company and her boss asks her to
find out what is the minimum the company can get away with in terms of an
affirmative action policy. She performs the same search as User 1 and gets
the same results back. However, User 2 is less than satisfied - she wanted
practical help, not a bunch of old articles and long-winded academic stuff.
As it happens there was some practical information on the site, but it was
swamped out by large numbers of journal article abstracts and archived
magazine articles, so she didn't find it.
Our proposed solution (I'm afraid ignoring the needs of one of the users is
not an option) is to add a metadata element to all content describing its
type (e.g. News Article, Journal Article, How-to-guide). Some of you may
equate this to the Dublin Core element: Type.
Then we plan to group search results in a simmilar way to Nielsen's Use It
site http://useit.mondosearch.com/cgi-bin/MsmFind.exe?QUERY=heuristics.
Our theory is that user's searching on the same keywords but for different
treatments of a subject can make an informed judgement based on samples of
results from each information type. So User 1 can see that the news archive
and journal articles are her best bet and drill down into those information
types, whilst User 2 can spot the How-to-guides because only the top 3 of
the other two (more numerous) types are displayed.
Assuming that we choose the right information types, can anyone see any
major flaws in this idea? For example, our solution will mean that results
will never be purely ranked by relevancy of search terms, but we're guessing
that will not be a major issue.
Can anyone point me in the direction of good literature on this subject?
TIA
PS By way of introduction: I work for a small agency in London as a project
manager/whatever-you-want-me-to-be-today
Jason
More information about the Sigia-l
mailing list