ISSUE-8: Use Case needed to represent rules in “immanent� ontologies

Use Case needed to represent rules in “immanent� ontologies

State:

CLOSED

Product:

(no longer used) Use Cases

Raised by:

Deborah Nichols

Opened on:

2006-08-15

Description:

Raised by: Bob Colomb. Posted to Issues by Deborah Nichols [DLN].

An external reviewer, Dr. Robert Colomb (colomb@itee.uq.edu.au), indicates
that there is a type of rules not covered in the UCs, viz., rules that are
generated dynamically within an application process by a statistical method.
His email includes references to this rule type within the OMG Ontology
Development Metamodel (ODM), along with a list of scenarios using such
rules.

Issue: Is this a missing category that the UC Document should cover? If so,
where in Phase I/II does it belong? Also, it would need characterization in
the RIF RAF. Would someone undertake to write a use case? If we decide not
to include this rule type, why not? How should we respond to Dr. Colomb’s
comment?

Detail: In email to public-rif-comments@w3.org, Bob Colomb raised the
following issue (original email archived at
http://lists.w3.org/Archives/Public/public-rif-comments/2006Apr/0001.html):

The use cases canvassed are all collections of rules generated by humans
outside the processes the rules are used to regulate. In the terminology of
the OMG Ontology Development Metamodel (ODM) submission, published also as
Hart et al (2004), the rules are all transcendent.

But there is an enormous effort in all sorts of areas to generate rules from
within the processes (immanent) by statistical means, many of which are
covered by techniques in the various subfields of data mining. These rules
are expressed in a variety of forms, but the forms are all
intertranslatable, and all equivalent to rule sets (Colomb 1999).

Some scenarios include:

The US government develops (by a contractor) a set of rules for face
recognition which is mandated for several government agencies and encouraged
for use by agencies of other levels of government and by various industries..

A telephone exchange manufacturer analyses usage patterns and minor fault
reports to arrive at a set of rules that can predict major failures. This
set of rules is used in the network management software by many telcos using
that equipment.

A pathology laboratory develops a set of rules for making clinical
interpretations of a class of blood test. These rules are exported to other
similar laboratories. This community of laboratories has a mechanism to
exchange improvements to the rule set. (Applies to a huge range of medical
diagnostic test artifacts.)

Security administrators at a major web site discover a new pattern of
traffic characterizing a new security threat. This pattern is formulated as
a ruleset and communicated widely among site security administrators.
(Applies to a wide variety of situations, including credit card fraud,
identity theft and money laundering.)

Dr. Robert M. Colomb
Reader
Data and Knowledge Engineering Group
School of Information Technology
and Electrical Engineering The most thought-provoking thing about
The University of Queensland our thought-provoking time
QLD 4072 Australia is that we are still not thinking..
www.itee.uq.edu.au/~colomb Heidegger
Phone +61 7 3365 1190
Fax +61 7 3365 4999


COLOMB, R.M. (1999) ³Representation of Propositional Expert Systems as
Partial Functions² Artificial Intelligence 109 pp. 187-209. See
http://www.itee.uq.edu.au/~colomb/Papers/PartialFunctions.html [Addresses
translation between production rules, decision tables, and decision trees.]

HART, Lewis, Patrick Emery, Robert Colomb, Kerry Raymond, Dan Chang, Yiming
Ye, Elisa Kendall & Mark Dutra (2004) ³Usage Scenarios and Goals For
Ontology Definition Metamodel² in Zhou, X., Su, S., Papazoglou, M.,
Orlowska, M. and Jeffery, K. (eds.) Web Information Systems Engineering
Conference (WISE¹04) 22-24 November, 2004, Brisbane, Australia. Springer
LNCS 3306 596-607. See http://www.omg.org/docs/ontology/04-01-01.pdf [Makes
the immanent/transcendent distinction wrt ontologies, including rules (see
discussions in Section 1.1 and 2.2; also Tables 1 and 2).]

[This was item 6 in:
http://www.w3.org/2005/rules/wg/wiki/UCR/Document_issues.]

Related Actions Items:

No related actions

Related emails:

No related emails

Related notes:

Display change log