Ontolog invited Speaker Presentation - Dr. Mark Greaves - Thu 2008.06.19    (1969)

Conference Call Details    (1HNB)

Attendees    (1J9E)

Agenda & Proceedings:    (1J9J)

Title: Knowledge Representation in Practice: Project Halo and the Semantic Web    (1J9L)

Vulcan's Project Halo is an ambitious, multiyear research program to develop a detailed scientific knowledge base that can answer AP-level questions and provide explanations in a user-appropriate manner. It is one of the larger AI research programs in the US today. Halo's current focus is building AI tools that allow graduate students in chemistry, biology, and physics to author scientific knowledge adequate to answer sophisticated natural-language questions without relying on trained knowledge engineers. Halo researchers have been working to link Semantic Web technology with the other knowledge representations in the system. This talk will lay out Halo's technologies and results to date, and describe the technical and UI issues we have faced in getting users to author scientific conceptual knowledge.    (1IKP)

http://ontolog.cim3.net/file/resource/presentation/MarkGreaves_20080619/MarkGreaves_20080619a.png    (1J9S)

                                    [Dr. Mark Greaves]    (1J9T)

Dr. Mark Greaves is currently Director, Knowledge Systems at Vulcan, Inc. Vulcan is the private investment vehicle for Paul Allen (co-founder of Microsoft). At Vulcan, he is sponsoring advanced research in large knowledge bases and advanced web technologies, including Project Halo.    (1J9V)

Formerly, Mark was Program Manager in DARPA's Information Exploitation Office (IXO) for the DAML, UltraLog, and Advanced Logistics Projects. At DARPA, he has sponsored research on logistics and supply chain control technologies, formal ontology specification, semantic web technology, and the application of software agent technology to problems of distributed control of complex systems-of-systems. Prior to coming to DARPA, he led advanced programs in software agent technology at the Mathematics and Computing Technology group of Boeing's Phantom Works division. His main research interests are in mathematical logic, semantic web, and software agent technology, about which he has published two books and several papers.    (1JB4)

Mark holds a Ph.D. from Stanford University, and can be reached at: <markg-at-vulcan.com>.    (1JB5)

Questions, Answers & Discourse:    (1JA0)

 PeterYim: Welcome to the Ontolog invited Speaker Presentation -
           Thu 2008.06.19 - 10:30am PDT / 1:30pm EDT / 17:30 UTC
           * Invited Speaker: Dr. Mark Greaves, Director, Knowledge Systems, Vulcan, Inc.  
           * Topic: "Knowledge Representation in Practice: Project Halo and the Semantic Web"    (1JM6)
 RaviSharma: Q for Dr. Greaves: Knowledge and understanding are intertwined, knowledge ought to be
             self managed! How far away are we from such intelligent concepts, contexts, etc. in 
             applications such as Search, Inference? are these available in Vulcan pilot in Chemistry
             or Physics examples?    (1JM7)
 RaviSharma: Q for Dr. Greaves: If you did not have NLP how did the SMEs input or interact with the
             Questions for Exams? VBisually or through Equations like in the Textbooks?
             How was that rendered?    (1JM8)
 SteveIhnen: It is hard to imagine learning - especially in biology and physics - without some form 
             of diagram understanding.  Is Project Halo working on this?    (1JM9)
 SteveIhnen: Following up on Ravi's question, Phase II did have NLP input.  Could you expand on what 
             technology was used for NLP input and give an example of a sample NLP question.    (1JMA)
 Brent Fegley: Dr. Greaves, thx for a most interesting presentation!  We know that knowledge bases 
               change over time -- even constants aren't static.   Does Halo have any provision for 
               relaying temporal or context-dependent knowledge in its answers -- appending to 
               past knowledge rather than overwriting it completely?    (1JMB)
 RaviSharma: Q for Dr. Greaves: Every commercially interesting software has similar debug history by 
              its user community such as databases, Operating and other applications including COTS. 
              What is the difference in Community based error corrections and how do you ensure that 
              SME's participate in correcting errors as these will correct them quickly?    (1JMC)
 JakubKotowski: Q for Dr. Greaves: could you elaborate a little more on the use of rules in Halo? 
                The rule language, syntax and what about the defaults and rules - there it may get 
                even more difficult for normal users.    (1JMD)
 Mike Bennett: So far, Halo seems to require the SME to learn and understand RDF/OWL concepts like 
               Object Property, Domain and Range. Is there any plan to be able to alias these to 
               plain English terms like Thing, Simple Fact, Relationship Fact? I want to be able to 
               gather knowledge from busy business SMEs without first training them in OWL.    (1JME)

Audio Recording of this Session    (1JAB)