Project Update: Feb. 5, 2002

Below is an OpenCyc project update in the form of a Q&A with Cycorp Founder Doug Lenat:

Q: Is Cycorp committed to releasing OpenCyc?

A: 100%. This has been in our plan since Cyc's inception in 1984. Two of the major uses of Cyc are (1) as a semantic kernel which others can extend, for their own projects and applications, and (2) as a sort of semantic glue or interlingua which others can use to align their ontology to and thereby interrelate their ontology with everyone else's which has similarly aligned with Cyc. OpenCyc is that kernel, that interlingua. We have several members of our technical staff working fulltime on this; our delay in releasing OpenCyc is in no way related to a lack of commitment or change in heart.

Q: What has caused the delays up to this point?

A: In April 2001 we expected we could just take the already-released Upper Cyc Ontology, update it, and release it within a couple months. Unfortunately, the deeper we got into that update, the more we realized that we'd come so far since its release (in 1997) that that sort of cosmetic update was not sufficient. We began a large-scale effort to re-do OpenCyc from the top down -- editing the full Cyc system and removing the "fringe", rather than by editing the small Upper Cyc Ontology. As that re-do effort proceeded, we also realized that a small percentage of what had been entered into Cyc over the past 17 years was poorly documented, incomplete, self-contradictory, or just wrong. A small percentage of 600 person-years worth of effort is still quite a large project -- to identify the problems and correct them -- and it has taken us the better part of a year to do it. The wait is worth it, however, and the result is something which will not have to be incompatibly modified after its release.

For most people, OpenCyc will be their very first exposure to Cyc technology. We felt it was more important to make that a good experience, to provide a quality product, than it was to meet a self-imposed release date announced by us in good faith but in error.

Q: What is the expected release date for OpenCyc?

A: Given the progress on the effort described above, we now expect Version 1.0 of OpenCyc to appear in July, 2002. In the meantime, we are far enough along that we plan to release version 0.6 next month -- during March, 2002. Version 0.6 comprises the full set of concept terms that will be in Version 1.0, but with only some of the axioms interrelating those terms which will appear in v1.0, and with a subset of the v1.0 power tools for editing, browsing, and running Cyc.
Even version 0.6 will be sufficiently large (about 5000 terms and 40,000 assertions) and functional (including a browser, editor, and inference engine) that users can start learning the Cyc ontology, representation language, and accompanying tools . Users can start aligning their applications to Cyc, can start using Cyc, even with version 0.6. Another way of stating this is that we are ready to have the OpenCyc community form and get started, next month.

Q: Do you still expect OpenCyc to have a major impact on the Web, on software development and on the world?

A: Of course. As we've said for almost 20 years now, something LIKE Cyc is needed for semantic sharing of knowledge across applications, to enable semantic integration to happen automatically (in contrast to n-squared labor-intensive post-hoc methods like data warehousing, and in contrast to the impossible requirement that the authors of the separate bodies of individual content work out detailed shared semantics ahead of time.) In modern terms, this means that something LIKE Cyc is needed for the Semantic Web. The delay of OpenCyc has only served to highlight the need for it. We're looking forward to finally filling that need.

Q: Is OpenCyc HAL?

A: OpenCyc is far from the complete AI system portrayed in 2001: A Space Odyssey, or in Star Trek. While Cyc itself continues to head squarely in that direction, OpenCyc 2002 is more of a framework to which others can align. As Cyc itself becomes more intelligent -- as its knowledge base grows and its reasoning system improves -- the benefits and power that come from such developments will be integrated into OpenCyc. This means that the growing intelligence that is Cyc will be more or less automatically utilized by those already using OpenCyc.

Posted by John De Oliveira 2002-02-05