A Knowledge -based model for representing and classifying companies
Loading...
Date
1989
Authors
Puvanesan, Premala
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The main purpose of t~is study is to develop a knowledge-based model for
representing ~nd classifying companies. It must be stated that to the best
of our knowledge, this is a novel ·undertaking. The model is focussed at
beina an exploratorv svstem and is not per se to produce a "real world" "" . .
model for classifying companies. In pursuit of this theme various
representation schemes, classification strategies as well as the
effectiveness of inheritance in such a task is explored.
To furnish ourselves with the problem-:solving know how of domain experts,
extensive knowledge engineering was diligently carTied out. We then went
on to review the major representation paradigms namely logic, rules,
semantic nets and frames which led to the choice of employing a
combination of frames· and logic via the knowledge engineering system,
FRAMELOG. This consequently led to the implementation of the model.
The main findings of the study show that frame structures amongst other
favourable features are suitable for ·declaratively representing data
· patterns pertaining to company categories. Ttle hierarcrlical organization of
company categories in the model enables the more specific categories to
inherit from their more general counterparts hence promoting a
compactness in th.e specification of rules. In the prototyping phase, we
realized that the classification strategy could be modelled in such a way
(through different frame declaration approaches) to exercise implicit or
explicit classification control. The strategy employing explicit control is
'
expected to be more viable when the need arises to incorporate more
complex conceptualizations in the knowledge-base. Besides automatically
classifying new companies, the system is also capable of reclassifying
companies (if necessary) when changes are made to the information
pertaining to the company in the knowledge base ..
The major result of -the knowledge-based model is in producing a
comprehensive list of comparable companies which fall within a similar
category. This is achieved by employing a generalized search method which
reduces the search space considerably by eliminating from consideration
candidate solutions that cannot succeed In a given situation. The task is
thus accomplished within· a reasonable space and time.
Description
Keywords
Representing , Classifying companies