IA Summit 2004 Taxonomies, Controlled Vocabularies, and Ontologies

Amy Warner, Katherine Bertolucci, Kathryn Lewellen
Program Description

These are my notes from their presentation on Feb. 28.

Katherine Bertolucci

use your client’s vocabulary but serve the needs of your client. (Story about selecting a different term that worked better for client’s system in the long-run.)

classification structures reflect our goals. should lead you to things that you didn’t know were there

Happiness is Taxonomy: Four Structures for Snoopy

Amy Warner

project lead on Z39.19 standard for controlled vocabularies (ANSI/NISO Z39.19, Guidelines for the Construction, Format, and Management of Monolingual Thesauri)

CDC

Testing and evaluation

exploratory card sorting

• asking members to name all the categories of a class or all the members of a category
• card sort testing whether people share an understanding of the meaning of terms
• name everything in the class. look for true hierarchical (is-a)

fixed card sort

• perform after main classes identified
• subjects assign labels or content to categories

Draft CV Review. Count on making revisions. Do things in an iterative way.

Kathryn Lewellen

Sr. Ontologist in Thomson's Legal and Regulatory division

Definitions

• AI/Academia Specification of a specialization
• Library - taxonomies
• Business. Why do we need a business definition?

RDF and Owl have yet to be implemented in a large biz

business

• fundamental shift from stabilizing structure to enhancing context and text
• API, Architecture (DTD/metadata)

benefits = combines human and automated efforts. generates revenue (ROI): ontology assets and services are proprietary. can be sold externally. ROI -- linking, alerts and precision searching. Metrics

Q&A

(I think I've attributed each answer to the correct panel member, but I was sitting on the floor and trying to identify them by voice...so don't count on it)

Audience member: methods for maintenance? Maintaining once they're built how do we keep them tuned?
Katherine Bertolucci: i design it the way they think and tell them it's a living thing
Amy Warner: I develop and leave but I help them develop a strategy, look at search logs then compare to their vocabulary to check for match. Run indexing logs to see how people are adding terms to the content to see how they are applying to terms to the content. there is a sort of inertia for revenue reasons and tendency to stay with what you have. Having formal processes can help you overcome that.

Audience member: How to translate to UI?
??: create drill down so each category level has a separate page.
Amy Warner: fish eye view -- where you are in the taxonomy is in larger text. Breadcrumbs
kathryn Lewellen: how for browsing, how for searching? drill down traditional hierarchy view may be ways to visual those.

Audience member: controlled vocabs to populate metadata fields
Amy Warner: consider those two things have to inform each other. content analysis, take to users and see if they were given this piece of content where would they place them. we love metadata fields but people have to fill in this descriptive metadata. So we need to figure out what the most important ones are to get the users (acting as indexers) to apply them when they fill them out.