RUDI users are currently testing a new search facility which
makes use of Dublin Core metadata.
This facility has been developed is response to user feeback which
suggested that improvements to our original search engine
(from Verity) were needed.
This search engine suffered from the common problem of uncertainty
about the relevance of results.
That is, when presented with a hit list, how does the
user know which query result is most likely to satisfy his/her
request for information? He/she can either narrow the scope by
constructing ever more complex boolean queries or browse the
results as returned. In the first approach, inexpert users
are often put off, and in the latter approach the scant and
often inappropriate information returned in a hit list requires
every result to be viewed.
RUDI is developing mechanisms which help to address this problem.
Metadata is being used to label basic HTML documents with,
for example, subject keywords, author names, and place names.
Each of these metadata fields can be used to focus a search, so that
the user can be much more confident of receiving a useful set of
results. The degree of relevance in this case is instilled in the data
by subject specialists, rather than relying on the vagaries of the
ranking algorithms built into most search engines.
Some features of the system are as follows:
* Metadata is stored with its corresponding document in plain
text files in the Dublin Core (DC) format.
* Metadata is served dynamically using server side includes
(the metadata can change without changing the document).
* Metadata is created and edited using web forms (similar to
DCdot).
* Full text and metadata is gathered and indexed using the
Netscape Catalog Server.
* Full text and metadata field specific searching can be
carried out using a simple or extended search interface.
http://rudi.herts.ac.uk/catsrch/catsrch.html
* Place name and subject keywords lists, which act as automatic
queries to the search engine, are provided. These will be
married with work on a controlled vocabulary for urban
design. http://rudi.herts.ac.uk/catsrch/placename.htm
Currently only a small part of the corpus has been addressed.
DC metadata will be added to the remainder as and when resources
become available.
Your comments would be welcome.
(I will not be able to reply before September 8).
Thanks for getting this far.
Dr W J Worthington
RUDI Project Manager
-----------------------------------------------------------------
-- Bill Worthington, ERDC, University of Hertfordshire, UK ----
------ email:[log in to unmask] tel:(0)1707 284166 -----
-------- http://rudi.herts.ac.uk/people.html#wjw ---------------
|