Cyc

Dateline: 02/19/00

Many of you are already familiar with the concepts of knowledge bases, ontologies, natural language interfaces, etc. This week we'll learn about Cyc, an immense, broad, multi-contextual knowledge base developed over the last fifteen years. The goal of this large-scale effort is to enable a multitude of knowledge-intensive products and services which will hopefully revolutionize the way in which people use and interact with computers.

The Cyc (as in en-cyc-lopedia) project was started by AI pioneer Douglas Lenat at MCC (Microelectronics and Computer Technology Corporation) in Austin, Texas as a research project in 1984. At that time the project was estimated to be completed in 10 years at a cost of approximately 50 million dollars. The approach used to capture common sense information for the project was to capture all of the knowledge - both implicit and explicit - in a hundred randomly selected articles in the Encyclopaedia Britannica. Lenat's method is characteristic for the branch of AI commonly named GOFAI - or Good-Old-Fashioned-Artificial Intelligence. It is also in direct contrast to the approach taken by another famous AI researcher and his former student, Rodney Brooks, who is developing embodied intelligent machines - take Cog for an example - mainly based on a sensory input and without complicated methods for planning and knowledge representation.

Going back to Cyc, the main idea is to create a knowledge base that would supply the basic knowledge needed to be applicable to many different applications. By building a sufficiently large knowledge base with this general knowledge, it is hoped that the it will be able to learn or create new inferences by itself. Contrary to the most expert systems available today that contain only the information they need to solve problems within a very narrow range of knowledge, Cyc is built upon a core of over 1,000,000 hand-entered assertions. As we will see, feeding this amount of information into a computer is a very tedious task. It involves equipping Cyc with the ability to cut through the layers of ambiguity inherent to the world we are living in. For example, the sentence "Napoleon died in 1821 - Wellington was saddened" required two months to enter into the knowledge base all the information needed to explain the concepts of life and death! This task of pumping accurate and up-to-date information into Cyc's knowledge base is assigned to the team of engineers, or "blue-collar philosophers," each of them experts in some academic field (linguistics, economics, philosophy, ...). Entry of information into the knowledge base can be done via an English-to-CycL html parser, one of Cycorp's interface tools.

The project was broken up into five essential components: the knowledge base, the CycL representation language, the inference engine, the interface tools and the application modules. The method of representation had to evolve over the years - while many other knowledge bases are built with frames but it was found that this did not scale well Cyc. Today it is a type of second-order predicate calculus, a trade-off between expressiveness (how easily you can say complicated things) and efficiency (how easily the machine can reason with what you've told it). The size of the database forced the Cyc's team to continuously innovate - for example, it was impossible to maintain one huge flat knowledge base, so it was separated into hundreds of contexts or microtheories. On the other hand, bottom-up examples from encyclopedias, newspapers, novels, etc. were gradually replaced by a more top-down methods, treating entire topics one at a time and in moderate detail. At the end of 1994, the Cyc program was mature enough to spin off from MCC as a new company, Cycorp, to commercialize the technology and begin its deployment.

The rise of Internet has and corporate intranets brought an entire new dimension to this story: Cycorp recently founded a new division called e-Cyc to bring the power of Cyc to these environments. The e-Cyc Portal Toolkit is a special release of the Cyc Knowledge Server that allows Internet and intranet web sites to add common-sense intelligence to their environments. It could be described as an advanced type of Internet search engine (unfortunately, seems that their test site is currently unreachable) that's being offered as a solution to the preponderance of time-consuming "false positives" and "false negatives" often produced in the process of browsing. e-Cyc separates out the different meanings of ambiguous concepts, making the search easy and offering more and more relevant results. Final version will introduce a "Teach Me" button on the screen - once users clicks on it, they'll be taken to a new page that will engage them in a dialogue that offers them the opportunity to explain to Cyc something that it doesn't know about the world. Now when everyone can become a "blue-collar philosopher" and participate in the programming of Cyc, the explosion in the amount of information being pumped into Cyc's knowledge base is expected, with the result of Cyc becoming exponentially smarter and smarter.

I have described only a fraction of possible applications of this technology. Other applications include semantic information retrieval, consistency-checking, deductive integration of heterogeneous data bases, natural language interfaces able to cope with realistic levels of ambiguity, intelligent brokering of goods and services, knowledge navigation, question-answering and more. For more information on computers that really understand the world of human beings, visit the Cycorp's Web site, or sites related to knowledge representation and ontologies.