Addressing cognitive issues in knowledge engineering
with Jambalaya
Neil A. Ernst, Margaret-Anne Storey,
Polly Allen
CHISEL research group
Computer Science Department
University of Victoria
Victoria, British Columbia, Canada
{nernst,mstorey,allenp}@uvic.ca
Mark Musen
Stanford Medical Informatics
School of Medicine
Stanford University
Palo Alto, CA
musen@smi.stanford.edu
ABSTRACT
Cognitive support in knowledge engineering is a growing concern,
and information visualization is a useful means to address this. We
identify some requirements for tools offering cognitive support, and
present a tool, Jambalaya, which addresses some of these concerns.
We identify some of its features and describe areas we are actively
improving.
1. INTRODUCTION
The field of knowledge capture and knowledge representation is
at a crossroads. The preceding years have been characterized by
ad-hoc development, and lately there has been a move towards a
more systematic approach to the development of knowledge-based
systems. The area of cognitive support in knowledge engineering
has also seen a lot of ad-hoc tool development. In this paper, we
demonstrate how our tool, Jambalaya, is evolving into an answer to
the challenges of providing cognitive support for knowledge engi-
neering. While the tool started without formal requirements analy-
sis, we have gone through several analytical methods to move it to
a more formal status.
In light of the Semantic Web initiative, the numerous ontologies
and systems it envisions to support machine-readable data online
will have a large impact on the knowledge engineering domain. We
believe that one of the byproducts of this success will be a require-
ment for much improved metaphors for understanding the ontolog-
ical commitment that these different ontologies make. Such cogni-
tive challenges are by no means new to this field, however. Stelzner
and Williams [19], for example, describe some of the concerns with
mapping between the user’s mental model of a system, and the sys-
tem model. In a tool designed to help students understand a med-
ical expert system related to MYCIN, GUIDON-WATCH, Richer
and Clancey [16] make the point that “. . . providing multiple views
of the same knowledge or behavior can help a user understand a
complex system”. From these early studies, it seems clear that new
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies
bear this notice and the full citation on the first page. To copy otherwise, or
republish, to post on servers or to redistribute to lists, requires prior specific
permission and/or a fee.
K-CAP’03, October 23-25, 2003, Sanibel Island, FL, USA.
Copyright 2003 ACM 1-58113-000-0/00/0000ã$5.00
systems are going to require well-designed interfaces to make the
Semantic Web vision workable. Well-designed implies that there
has been some form of iterative, systematic, user-centered design
methodology involved in creating the tool. This paper makes the
case that such an interface ought to consider, if not actively involve,
techniques and tools from information visualization. In the remain-
der of this paper, we explain what information visualization is, and
describe some other approaches to visualization in knowledge en-
gineering. We then detail a tool we have contributed, Jambalaya,
beginning with its history, some studies we performed to analyze
requirements for the tool, and concluding with a detailed descrip-
tion of some of the functionality using a sample ontology. Finally,
we show how this tool addresses some of the concerns raised above.
2. BACKGROUND
Information visualization techniques are used in many domains to
help provide insight or to communicate information [5]. Infor-
mation visualization leverages innate human abilities to perform
spatial reasoning and make sense of relatively complex data us-
ing some form of graphical representation language. In the do-
main of knowledge engineering, such a language is often based
on graph theory and has two components: one, the use of nodes
to represent concepts in a domain; the other, the use of edges to
represent relationships between concepts. The language for visu-
alizing information in this domain therefore consists of manipu-
lations of graphs in some form or another. Jambalaya is closely
integrated with Prote´ge´ [12], an ontology engineering and knowl-
edge acquisition tool created at Stanford University. Prote´ge´ uses
a frame-based knowledge representation to allow users to model
domains using classes (concepts), instances, slots (relations), and
facets (constraints on the slots). Written in Java, its architecture al-
lows for extensions to be added via a plug-in metaphor. Recently,
work has been ongoing to make the tool compatible with the OWL
ontology language for the Semantic Web, as well as support web-
specific concepts such as namespaces and Universal Resource In-
dicators (URIs). Fig. 1 shows the default classes view with a slot
window open. The example used here, and throughout this paper,
is the wines ontology.
3. RELATED WORK
Earlier work was motivated by a realization that few users could
easily understand what an expert system tool was doing or how to
make it work. Concept maps [11] were more focused on user sup-
port than other efforts, but seemed to have been hamstrung by a lack
of supporting environments for easy knowledge acquisition. An-