OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

topicmaps-comment message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]


Subject: [topicmaps-comment] Topic Map or not for a taxonomy


Hi

Jumping (back) into another topic, triggered by an interesting article named
"A Taxi in Knowledge Land"
by Thomas Bandholz some weeks ago, ref:
http://195.14.240.28/123/03-05-03.html

I found that story very inspiring, since I happen to work on a similar
project.
We are designing a data structure (and the process) for taxonomies to
drive quite a fat web portal. So far I haven't found too much information
about
similar projects. I kind of wanted to share my thoughts and findings so far
and
looking forward any comments, thanks.

Our client has a typical situation where a wildly grown farm of web sites
has gone pretty much out of control in terms of classifying content and
managing
overall navigation structures - the number of web documents (either stored
locally or
referring external ones) is closer to 60'000 so far =>
they are about drowning in infoglut as some may say...

So, the approach has been that: let's get an automated Content
Classification (CC)
tool to harvest what is in there. Then by using partially human assisted
process to categorize
documents and dropping them to applicable categories (topics). All this is a
by-plot in the bigger portal project, (which doesn't necessarily help when
trying to look after designing a sound information architecture :|

I am just wondering if anybody has experience on integrating CC tools (such
as
www.quiver.com or http://www.inxight.com or what ever) and XTM
technology?...

To me CC and XTM are supplementary technologies,meaning automated topic map
generation to
provide taxonomies for a portal application, but I am learning that CC apps
do not talk XTM.

Another question rose when I started to figure out
what might be the ideal method/notation for storing the taxonomy.

So far I found following alternative paths:

-Custom built Taxonomy XML schema, requires more competence from all the
parties, more emphasis on long term benefits, can be tailored for ideal
match for information and process requirements, could be quite similar to
"Topic Map Friendly W3C schema" as introduced in the Taxi report. Importing
results from Classification tool and import/export to the Portal tool may be
easier to provide using this approach.

-Resource Description Framework (RDF), resource centric information
management technology, can be used for describing topic hierarchies, well -
are there any mature tools for editing and processing Taxonomy RDF
instances,
is anybody using RDF other than Open Directory? And it's missing the
merging.

-RDF Rich Site Summary (RSS see:
http://groups.yahoo.com/group/rss-dev/files/specification.html) ,an RDF
application, which is generally used for any situation when a
machine-readable list of textual items and/or metadata about them needs to
be distributed. There is even a taxonomy module draft available - but is
anybody really using this either?

-XML Topic maps,+++ , but I'm not sure if provides ideal support for
processing documents, for example,
facets are missing compared to from ISO model, which I originally thought
using to store arbitrary metadata. Powerful concept of Associations, but
leads to frequent look-ups to find out which sub topics are contained under
certain topic (child-parent relationship). TMCL is also in the wish list -
yet to come, so we may follow "Taxi's mpath approach".

Then drafting a bit the process:

Possible CC Scenario using Topic Maps (simplified)

1. Initial Topic hierarchy - taxonomy - are defined as Topics in XTM
instance.

2. Documents are classified by a Content Classification tool.

3. URLs pointing at each classified document are stored in Occurrence
elements
   under the matching topic(s) in the taxonomy XTM instance

4. InstanceOf elements can be used to bind topics to pre-defined Taxonomy
Object Classes
   as defined in the Taxonomy Topic Map template.

5. Scope feature of the XTM is used for attaching 2nd language names on each
topic.

6. In editorial process subject matter experts validate the final taxonomy
   using Topic Map editing tools guided by some sort of constrain rules.

7. Taxonomies get published via portal application, which reads the Topic
Map instance(s)
   and visualizes navigation structures to the end users.

Note
   Portal user can pick or create a user profile, which is then matched to
suitable
   Scope to filter out a personalized view on portal information.

   PSIs could be used to define a global site thesaurus available for
localized topic names to refer to.


Heimo Hanninen
XIA Information Architects Corporation

6 Gurdwara Road, Suite 205
Ottawa, Ontario,
Canada K2E 8A3

Tel: (613) 723 7759 #224
Fax: (613) 723 0321
Email: hjh@xiacorp.com



[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]


Powered by eList eXpress LLC