Ontolog Invited Speaker Presentation - Dr. Benjamin Grosof - Thu 2013.06.20    (3TW8)

Session Chair: Dr. LeoObrst (ONTOLOG; MITRE)    (3TWB)

Invited Speaker: Dr. BenjaminGrosof (Benjamin Grosof & Associates; Coherent Knowledge Systems)    (3TWC)

Title of Presentation: "Rulelog: Highly Expressive yet Scalable Logic enabling Cost-Efficient Ontologies"    (3TWD)

Archive:    (3U39)

Conference Call Details    (3TWE)

Attendees    (3TXB)

Agenda & Proceedings    (3U27)

Abstract:    (3U2F)

"Rulelog: Highly Expressive yet Scalable Logic enabling Cost-Efficient Ontologies" ... [ [ slides] ]    (3U2G)

http://ontolog.cim3.net/file/resource/presentation/BenjaminGrosof-Rulelog_20130620/BenjaminGrosof_20130620b.jpg [ Dr. Benjamin Grosof ]    (3U2H)

We present Rulelog, a new logic that is highly expressive yet scalable, and Textual Logic, a related new approach to natural language processing that does logic-based mapping between English text and Rulelog knowledge. We have developed their theory and implementation techniques in over a decade of research work, including in Vulcan's SILK and Project Halo, jointly with many others. A commercial implementation is underway from Coherent Knowledge Systems, a new startup. RIF-Rulelog is in draft as an industry standard submission to W3C from RuleML.    (3U2I)

Rulelog extends normal declarative logic programs (LP) to permit defeasible higher-order logic formulas but achieves computational tractability. Two keys to that tractability are: avoiding unrestricted reasoning-by-cases; and restraint, a kind of bounded rationality that leverages the third truth value ("undefined") in the well founded semantics. Flora-2, an open-source semantic rule engine (built on top of open-source XSB Prolog), implements most of Rulelog's expressiveness.    (3U2J)

Rulelog is a "big tent" knowledge representation (KR) that is sound semantically relative to classical logic (e.g., OWL-DL and Common Logic), in addition to supersuming LP (e.g., SPARQL, SQL, and essentially RIF-BLD, OWL-RL/RDF(S), and semantic production rules). Knowledge can thus be translated semantically into Rulelog from all of the above, typically without major (or, in many cases, any) practical information loss.    (3U2K)

Rulelog's expressive power enables cost-efficient knowledge acquisition (KA), including about ontologies. KA can start from effectively unrestricted English, via Textual Logic, which includes:    (3U2L)

1. Rapid interactive logical disambiguation: Effectively unrestricted English sentences can be semi-automatically disambiguated into Rulelog in rapid fashion. In a pilot study in the domain of college-level biology, roughly 10 minutes of labor per sentence was required by knowledge engineers, using Automata Linguist, a new tool.    (3U2M)

2. Textual terminology: Any English phrase corresponds one-to-one to a logical term. Ontological terminology emerges naturally and automatically from English text's phraseology, instead of requiring laborious separate specification and social agreement. This leverages background lexical ontological knowledge about English, available open-source.    (3U2N)

In addition, Rulelog's expressive power enables:    (3U2O)

3. Seamless mapping: Ontology mappings can be represented concisely and flexibly, and tightly combined in the same logic/system/knowledge-base with ontologies and rules/facts knowledge that employs the ontologies' terminology.    (3U2P)

4. Inheritance with exceptions: Ontologies based on class hierarchies can be made more concise and reusable by allowing inheritance to have exceptions, rather than requiring inheritance to be strict (as in OWL/RDF(S)).    (3U2Q)

Additional Resources    (3UPD)

About the Speakers    (3U2R)

Benjamin Grosof┬áis an industry leader in knowledge representation, reasoning, and acquisition. He has pioneered semantic technology and industry standards for rules, the combination of rules with ontologies, the applications of rules in e-commerce and policies, and the acquisition of rules and ontologies from natural language (NL). He has had driving roles in RuleML, W3C RIF (Rule Interchange Format), and W3C OWL-RL (rule-based ontologies). He led the invention of several fundamental technical advances in knowledge representation, including courteous defeasibility, restraint bounded rationality, and the rule-based technique which rapidly became the currently dominant approach to commercial implementation of OWL. He has extensive experience in machine learning, probabilistic reasoning, and user interaction design.    (3U2S)

Dr. Grosof has experience applying core technology for knowledge, reasoning, and related HCI in a wide variety of application areas, including: trust/privacy/security, contracts, compliance, legal, and services engineering; financial/insurance services, risk management, and regulations; defense and national intelligence; biomedical research; and data/decision analytics. From fall 2007 to early 2013, he led a large research program in Artificial Intelligence (AI) and rule-based semantic technologies at Vulcan Inc. for Paul G. Allen; this centered around the SILK system for highly expressive, yet scalable, rules. Previously he was an IT professor at MIT Sloan (2000-2007) and a senior software scientist at IBM Research (1988-2000). He is president of the expert consulting firm Benjamin Grosof & Associates founded while he was at MIT, and is affiliate faculty in the AI group at U. Washington. His background includes 4 major industry software releases, 2 years in software startups, a Stanford PhD (Computer Science), a Harvard BA (Applied Mathematics), 2 patents, and over 50 refereed publications.    (3U2T)

Transcript of the online chat during the session:    (3U2U)

 see raw transcript here.    (3U2V)
 (for better clarity, the version below is a re-organized and lightly edited chat-transcript.)
 Participants are welcome to make light edits to their own contributions as they see fit.    (3U2W)
    -- begin in-session chat-transcript --    (3U37)
	------
	Chat transcript from room: ontolog_20130620
	2013-06-20 GMT-08:00 [PDT]
	------    (3UPY)
	[9:15] PeterYim: Welcome to the    (3UPZ)
	 = Ontolog Invited Speaker Presentation - Dr. Benjamin Grosof - Thu 2013.06.20 =    (3UQ0)
	* Session Chair: Dr. LeoObrst (ONTOLOG; MITRE)    (3UQ1)
	* Invited Speaker: Dr. BenjaminGrosof (Benjamin Grosof & Associates; Coherent Knowledge Systems)    (3UQ2)
	* Title of Presentation: "Rulelog: Highly Expressive yet Scalable Logic enabling Cost-Efficient Ontologies"    (3UQ3)
	Logistics:    (3UQ4)
	* Refer to details on session page at: http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2013_06_20    (3UQ5)
	* (if you haven't already done so) please click on "settings" (top center) and morph from "anonymous" to your RealName (in WikiWord format)    (3UQ6)
	* Mute control: *7 to un-mute ... *6 to mute    (3UQ7)
	* Can't find Skype Dial pad?
	** for Windows Skype users: it's under the "Call" dropdown menu as "Show Dial pad"
	** for Linux Skype users: please note that the dial-pad is only available on v4.1 (or later or the earlier Skype versions 2.x,)
	   if the dialpad button is not shown in the call window you need to press the "d" hotkey to enable it.    (3UQ8)
	Attendees: AdrianPaschke, AlexShkotin, AmandaVizedom, BenjaminGrosof, BobSmith, BobbinTeegarden, 
	BrianHaugh, ElieAbiLahoud, HaroldBoley, HensonGraves, JamesOdell, JoelBender, JulienCorman, 
	KenBaclawski, LamarHenderson, LeoObrst, MichaelBarnett, NaicongLi, OrisFriesen, PaulenaMeyer, 
	PlamenPetrov, PatrickMaroney, PeterYim, SteveRay, TaraAthan, ToddSchneider, TonyWeida, bryan    (3UQ9)
	 == Proceedings: ==    (3UQA)
	[9:22] anonymous morphed into BrianHaugh    (3UQB)
	[9:27] anonymous1 morphed into JamesOdell    (3UQC)
	[9:28] anonymous2 morphed into OrisFriesen    (3UQD)
	[9:28] anonymous morphed into PaulenaMeyer    (3UQE)
	[9:29] anonymous morphed into TonyWeida    (3UQF)
	[9:29] anonymous1 morphed into AlexShkotin    (3UQG)
	[9:31] anonymous morphed into AdrianPaschke    (3UQH)
	[9:32] anonymous morphed into ElieAbiLahoud    (3UQI)
	[9:33] anonymous1 morphed into NaicongLi    (3UQJ)
	[9:33] anonymous2 morphed into TaraAthan    (3UQK)
	[9:33] anonymous1 morphed into JulienCorman    (3UQL)
	[9:34] anonymous morphed into LamarHenderson    (3UQM)
	[9:35] anonymous morphed into PatrickMaroney    (3UQN)
	[9:35] AlexShkotin: How to download slides?    (3UQO)
	[9:36] HaroldBoley / PatrickMaroney: Slides 
	at: http://ontolog.cim3.net/file/resource/presentation/BenjaminGrosof-Rulelog_20130620/Rulelog--BenjaminGrosof_20130620.pdf    (3UQP)
	[9:36] AlexShkotin: Thank you!    (3UQQ)
	[9:36] PeterYim: == LeoObrst starts the session ...    (3UQR)
	[9:37] anonymous morphed into PPetrov    (3UQS)
	[9:39] PeterYim: == BenjaminGrosof starts his presentation ... slides at: http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2013_06_20#nid3U3C    (3UQT)
	[10:20] anonymous morphed into bryan    (3UQU)
	[10:20] LeoObrst: It sounds like "restraint" + avoidance of reasoning-by-cases is a kind of 
	higher-level guard on the reasoning that gets performed: is that right?    (3UQV)
	[10:33] AmandaVizedom: (ref. slide 7) I'm puzzled by the idea of 1-1 relationship between logical 
	constants and "words/word-senses", since words and word-senses do not have a 1-1 relation. Also this 
	seems to create enormous barriers to cross-lingual reasoning, or even use of K resources... Unless 
	this is a K *layer*? That is, are you talking about a direct path from NL to KR using this "textual 
	logic" approach, then further connections between the resulting KR and more language-independent KR?    (3UQW)
	[10:37] LeoObrst: On slide 7, you say that there is a "natural composition" of textual phrase <--> 
	logical term. But it seems like the entire phrase is just chunked, rather than the result of 
	composition, i.e., the result is not a formula.    (3UQX)
	[10:41] SteveRay: On slide 15, it looks like the transformation lost the constraint of the 
	"hydrophobic" subclass of possibly all kinds of tails...    (3UQY)
	[10:42] SteveRay: ...at least in the ""formula"" section of the UI, but not in the ""Readings"" section.    (3UQZ)
	[11:18] HaroldBoley: On slide 15, the chunked phrase on the top line, 'the'('hydrophobic'(...)), is 
	to be transformed into the central yellow-box formula Forall(?x9)...    (3UR0)
	[10:44] AmandaVizedom: +1 for representation of queries in same language as knowledge.    (3UR1)
	[10:53] LeoObrst: What about intensional contexts in natural language (e.g., propositional 
	attitudes)? Example: John believed/considered/etc. that the conclusions were wrong.    (3UR2)
	[11:01] PeterYim: == we're in the Q&A (and general discussion) segment now ...    (3UR3)
	[11:05] AmandaVizedom: Have to drop off. Thanks, Benjamin, very interesting.    (3UR4)
	[11:06] PatrickMaroney: Flora-2 tutorial: http://coherentknowledge.com/wp-content/uploads/2013/05/Flora-user-tutorial-main.pdf    (3UR5)
	[11:12] PeterYim: the OntologyBasedStandards effort that we are after will benefit hugely from using 
	textual logic that you talked about today ... we'll look forward to working with you and your 
	colleagues, Benjamin - ref. http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologyBasedStandards    (3UR6)
	[11:20] PeterYim: question from BrianHaugh ref. SILK and its relations to Flora ... Semantic 
	Inferencing on Large Knowledge (SILK) knowledge representation system - http://silk.semwebcentral.org/    (3UR7)
	[11:20] ElieAbiLahoud: I have to drop off. Thank you Benjamin.    (3UR8)
	[11:29] PeterYim: great talk ... thank you, Benjamin ... thanks everyone!    (3UR9)
	[11:29] LeoObrst: Great talk, Benjamin! Thanks all, for participating.    (3URA)
	[11:29] PeterYim: join us again, same time next week, for the Ontolog Invited Speaker Presentation 
	by Dr. VinayChaudhri of SRI - see session details at: http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2013_06_27    (3URB)
	[11:30] PatrickMaroney: Some additional related resources: http://haleyai.com/wordpress/    (3URC)
	[11:30] PeterYim: -- session ended: 11:30am PDT --    (3URD)
    -- end of in-session chat-transcript --    (3U38)

... More Questions?    (3U2Y)


For the record ...    (3U31)

How To Join (while the session is in progress)    (3U32)