Automating Knowledge Acquisition for Expert Systems by Sandra Marcus (auth.), Sandra Marcus (eds.)

By Sandra Marcus (auth.), Sandra Marcus (eds.)

In June of 1983, our professional platforms learn team at Carnegie Mellon college started to paintings actively on automating wisdom acquisition for professional structures. within the final 5 years, now we have built numerous instruments less than the strain and effect of establishing specialist structures for company and undefined. those instruments comprise the 5 defined in chapters 2 via 6 -­ extra, MOLE, SALT, KNACK and SIZZLE. One test, performed together by means of builders at electronic apparatus company, the jump learn team at Carnegie Mellon, and contributors of our crew, explored automation of information acquisition and code improvement for XCON (also often called R1), a production-level specialist method for configuring DEC computers. This paintings encouraged the advance of RIME, a programming method built at electronic that's the topic of bankruptcy 7. This e-book describes the foundations that guided our paintings, seems intimately on the layout and operation of every instrument or technique, and studies a few classes discovered from the company. of the paintings, introduced out within the introductory bankruptcy, is a standard topic that a lot strength could be won by means of figuring out the jobs that area wisdom performs in challenge fixing. every one software can take advantage of such an realizing since it makes a speciality of a good outlined problem-solving procedure utilized by the professional platforms it builds. each one instrument bankruptcy describes the fundamental problem-solving process assumed by way of the software and the leverage supplied by way of committing to the method.

Show description

Read Online or Download Automating Knowledge Acquisition for Expert Systems PDF

Best nonfiction_8 books

Parallel Processing for Jet Engine Control

Parallel Processing functions for Jet Engine regulate is a quantity within the new Advances in business regulate sequence, edited via Professor M. J. Grimble and Dr. M. A. Johnson of the economic regulate Unit, college of Strathclyde. The publication describes the mapping and cargo balancing of gasoline turbine engine and controller simulations onto arrays of transputers.

Variational Methods for Discontinuous Structures: Applications to image segmentation, continuum mechanics, homogenization Villa Olmo, Como, 8–10 September 1994

Lately many researchers in fabric technological know-how have concentrated their consciousness at the examine of composite fabrics, equilibrium of crystals and crack distribution in continua topic to quite a bit. even as numerous new concerns in desktop imaginative and prescient and snapshot processing were studied extensive. the knowledge of a lot of those difficulties has made major growth due to new tools constructed in calculus of diversifications, geometric degree idea and partial differential equations.

Coping with Financial Fragility and Systemic Risk

Dealing with monetary Fragility and Systemic hazard identifies and discusses the assets of perceived fragility in monetary associations and markets and its strength effects through the economic climate. It then examines inner most area recommendations for facing systemic probability and mitigating the results.

Forest site and productivity

An information of wooded area web site and woodland productiveness variables is prime to sound wooded area perform in every single place. the power to spot websites and placement difficulties competently and manage productiveness variables for upkeep or development of productiveness is the root of recent woodland administration. even supposing the elemental evidence concerning woodland website and productiveness observe through the international, the applying of knowledge and the reaction to manipulation differ enormously and rely on neighborhood woodland stipulations.

Additional resources for Automating Knowledge Acquisition for Expert Systems

Sample text

At the same time, however, it raised several new problems. Most significant were issues of control, knowledge-base representation, user interface, and user-dependency. Subsequent systems, MOLE and TDE, address these problems. Acknowledgements MORE was developed in collaboration with John McDermott and Steve Nowlan while the author was at Carnegie Mellon University. TDE and TEST were developed at Carnegie Group, Inc. in collaboration with Edwin H. Breaux, Rajiv Enand, Peter DeKlerk, Robert L. Joseph, AI Kepner, Jeff Pepper, and Bill Richer.

MORE's Problems As a research prototype, MORE demonstrated the feasibility of our approach. At the same time, it raised several new problems that became critical concerns. Of most importance were issues of control, knowledgebase representation, user interface, and user dependency. The essential control issue for MORE was knowing when to ask questions. Like the successful knowledge engineer, knowledge-acquisition systems need to not only ask good questions but also conform to the domain expert's desire to provide information as it comes to mind or to skip questions whose MORE: From Observing Knowledge Engineers 31 answers are unknown.

Because frequency conditions affect the interpretation of all the symptomatic evidence associated with a hypothesis, these rules may be applied whenever the evidential effect of any symptom associated with the conditionalized hypothesis is appraised during a diagnostic session. All hypothesis-condition rules bearing on the same hypothesis are considered to be members of the same rule family. As rules are created, MORE asks the knowledge-base developer, or expert, to assign positive and negative confidence factors to them.

Download PDF sample

Rated 4.30 of 5 – based on 7 votes