Patrick Cassidy (35YC)
Ontologist; president of MICRA Inc.; consultant on ontological engineering. (35XL)
[picture of Dr. Patrick Cassidy] (O8P)
Patrick Cassidy MICRA, Inc. 735 Belvidere Ave. Plainfield, NJ 07062-2054 cassidy@micra.com (908) 561-3416 Cell: 908-565-4053 (3TRS)
Background (35XO)
- Ph.D. (Chemistry) Harvard University, 1971 (35Y6)
- Biochemical and microbiological research at Merck & Co. 1967-1993 (35Y7)
- Ontologist at VerticalNet 2000-2001 (35Y8)
- Sr. AI Engineer, MITRE Corp. 2004-2007 (35Y9)
- President of MICRA Inc., consultant on ontological engineering (3TRT)
- Involved in building ontologies since 1992. (35XP)
Personal Objectives (35XQ)
(1) To help advance construction of a reference upper ontology that could serve as the logical defining vocabulary for any domain or specific application -- to promote efficient re-use of research results and semantic interoperabilty. (35XR)
(2) To help create a language understanding program that can approach human-level language ability for the most basic levels of language, such as that achieved by 5 or 6-year olds. (35XS)
Areas of Interest (35XT)
My ongoing interest is in the development of natural language understanding systems. I believe that the development of an ontology that can serve to define the concepts in a computational lexicon is a necessary preliminary to the problem of achieving human-level language understanding. Several good ontologies are available, but some level of broad agreement on one upper ontology appears to be a necessary preliminary to the efficient conduct of such research by promoting re-use of results among multiple research groups. One personal goal is to work to develop such agreement among some community sufficiently large to provide the networking benefits of multiple approaches to a common problem. (35XU)
It also seems likely that the development of a widely-used upper ontology will require substantial funding. The best mechanism to obtain funding for such a project is still an active topic of discussion. (35XV)
Topics for Discussion (35XW)
There has been ample discussion of formats for ontology description in other fora; I hope that any discussion of formats will not be redundant with those discussions. What I think would be valuable at this point are: (35XX)
- 1. Accumulation of a set of examples where ontologies have been used in practical applications, with some evaluation that could allow an estimate of how much money was saved by the use of the ontology as compared with alternative resources for data storage (e .g. a traditional database). This could provide the evidence of benefit necessary to convince potential users to adopt ontologies, and potential funders to allot more money for research. (35Y4)
- 2. Maintaining a list of applications that use ontologies, are available to the public, and could serve as test programs to evaluate alternative ontological theories. (35Y5)
- 3. An effort at direct construction of a consensus upper ontology (called the COSMO) by a volunteer discussion group is already underway within the ONTACWG. This effort requires that the participants suggest those highest-level concepts that they feel are necessary to represent their domain concepts, and then discussion could focus on the question of whether those proposed upper-level concepts are logically contradictory, or could in fact be accommodated within a single logically consistent ontology (which may have some redundancy). Where logically contradictory representations appear necessary, the next question would be how to include such alternative theories while minimizing the differences with other ontologies. This would help us to learn how to maximize interoperability even when logically contradictory representations are desired by different developers. (35XY)
This could help answer an important question: whether the differences in existing ontologies are due to a true need for logically contradictory representations, or in fact arise as an accident of independent development and individual preferences of the developers. A related question is whether one representation may be preferred because it allows more efficient computation than another, rather than being uniquely necessary for representation. Having example applications that could be tested with different ontologies would help to resolve such questions. (35XZ)
The starting ontologies for construction of the COSMO include OpenCyc, SUMO, DOLCE, BFO, ISO15926, and the GUM, but any others that supply general concepts not already present in any of the starting ontologies will be very welcome and useful. (35Y0)
In coordination with the development of the COSMO, I am also trying to promote the use of a restricted defining vocabulary of English to be used as the vocabulary for defining concepts in any knowledge classification system. Such a linguistic defining vocabulary should eventually be mapped to the COSMO so that those who want to build ontology-based knowledge bases can do so using linguistic definitions and queries without the need to learn the details of ontological knowledge represention. (35Y1)
The latest version of the COSMO ontology is now maintained at the MICRA web site. (3VTD)
The history of the COSMO project is here: Wiki Refresh for the Ontology and Taxonomy Coordinating WG. See new Wiki pages. (3VNR)