Actions

Ontolog Forum

The printable version is no longer supported and may have rendering errors. Please update your browser bookmarks and please use the default browser print function instead.

Ontology Summit 2018 Summary

The OntologySummit is an annual series of events that involves the ontology community and communities related to each year's theme chosen for the summit. The Ontology Summit was started by Ontolog and NIST, and the program has been co-organized by Ontolog, NIST, NCOR, NCBO, IAOA, NCO_NITRD along with the co-sponsorship of other organizations that are supportive of the Summit goals and objectives.

Description

The Summit theme this year is Ontologies in Context.

In general, a context is defined to be the circumstances that form the setting for an event, statement, or idea, and in terms of which it can be fully understood and assessed. Some examples of synonyms include circumstances, conditions, factors, state of affairs, situation, background, scene, setting, and frame of reference. There are many meanings of "context" in general, and also for ontologies in particular. The summit this year will survey these meanings and identify the research problems that must be solved so that contexts can succeed in achieving the full understanding and assessment of an ontology.

  • In the agent world, the closest approximation to context is a BDI (Belief, Desire, Intent) agent, primarily expressed using modal operators.
  • Other examples of context include
    • Speech act theory
    • Situation semantics
    • Kamp/Heim Discourse Representation Theory (DRT)/file-change semantics
    • Dynamic semantics
  • There are several ways to represent contexts within a logical framework, including:
    • Languages such as IKL (IKRIS) have a context logic built into them.
    • Hybrid logic can also be used to specify contexts.

Session Summaries

  • Sessions in 2017 and 2018 are being summarized based on presentations, recordings and chats. Draft summaries will not be posted until summary for each session is completed (In Progress)
  • Dr. Ravi Sharma (Editor)

Summary of 2017-09-13 First Research Session

  • Dr. Ravi Sharma (Editor)
  • Ken Baclawski summarized Pat Hayes Paper “Contexts In Context” published around 1997. Described several concepts from the paper. External Language (EL) and Internal Coding (representation) IC. Single words require context, and sentence parsing as well. Ken described 4 different contexts notions - Physical, Linguistic, Conceptual and Deductive. Physical or topic is about objects e.g. about the environment, Linguistic is more like programming stack (that allows switching) about roses being described in garden, are not they beautiful, distinction between them is often blurred. Conceptual is background describing context e.g. bank (of a river or place for depositing cash). It does not change so rapidly and banking of airplane or edge of river are examples from general knowledge. Cognitive semantics goes beyond this than linguistic such as parsing sentence. Exposure of meaning, deductive is beyond complete description of time context. That Deductive context framework is inadequate.
  • John Sowa, Pat Hayes worked together for years and in a conference in 1999 on Context, Guha gave a talk with McCarthy, discussed huge variety and complexity in linguistics and NL parsing. For every relation there are ten other, extra argument can be context but very complex. Logic with Meta Language so that you can include reasoning with background knowledge and we developed IKL.
  • “That” is the keyword additionally used in IKL. Another way is to Quote and thus take multiple sentences with an AND. In a sense a simple addition to CL thus creates meta language. Ist (it-is true) is definable in CL related to “that” operator for a block of statements and IKL is useful. Such blocks can be evaluated, John offered to give a talk, and said 2004-2006 IKL funding was reduced but never funded further. Pat Hayes, Chris Menzel and John Sowa worked on IKL revival and still there is hope. Matthew West referenced revival, integration would make it explicit.
  • John Sowa and David Whitten conversed on the evolution of IKL as result of 2-year IKRIS Project effort in which Leo Obrst also provided many context related definitions. IKL for example uses “that” to provide Quoted forms and ist (it-is-true) logic. Readers can also refer to subsequent long postings on Ontolog Forum on topics, such as, IKL to CL, CYC etc. Situation semantics, “that” operator, with micro-theory could combine CL with “That” operator acting as context. CL + “That” Operator and Language processors and modules gave flexibility to handle Language, reasoning, powerful mechanism. CL Brennen implemented CL and “that” added is IKL.
  • David Whitten posed many questions “context in ontology development and reuse” e.g. Montague and semantics, and IC representation in global variables to be reused based on context, described above 4 types of contexts and nicely summarized this session in chat statements available on this session page.
  • Context based first order logic is proposed by Hays-referenced authors Guha, Buvac and others, in which logic can be presented in given context and inferences are provided for moving between contexts.
  • Todd Schneider and Ram Sriram provided their inputs on how to go about the topic of the Summit and steps required to plan sessions and discover authors and research work on the topic. Ram Sriram: Understanding what Context is in our Context, Ken responded what is most relevant. Todd expressed, context is to be understood while developing or reusing ontologies.
  • Ken: Leo and Nichols paper referenced on Session pages is about 12 years old. It is less clear but with AI perspective, 2 or more contexts could be used and then reasoned about, using micro-theory ensuring compatibility. Leo looks at annotations and reifications as mechanisms to represent context but without semantics. Security classifications reasoning was discussed and John Sowa commented that with meta level operations outer context can have reasoning and modules can be handled with Quotation operation.
  • Ken Discussed last paper by Blackburn but from our point of view we need to understand how it is relevant. John Sowa said flattening makes it messy with pointers all over the place for context. Tarski style model theory or simplicity for the feature of logic that is implied. Ken described aspects of hybrid logic, and modal logic. John said reference to McCarthy’s work - every possible world - replace with modal logic and distinction for necessary and true and ontic logic, temporal logic (always, sometime) and use of “that” operator based logic. Hybrid and Modal logic are related. Language and Logic synonyms. Formalized languages have logic built in the language. Back Quote in LISP and “that” operators have similarities.
  • Matthew West: Barry Smith and Matthew West prefer making a top-level ontology. Another approach - Micro-theory approach with local theories that work independent of each other.

2017-09-20 Second Research Session

  • Presented by John Sowa, Reasoning with and about contexts: the 'ist' and 'that' operator ==
  • Slide 1 – Definitions, first context in NL, any text, from physical or oral info describing text, NL flexible hard to process by computer, second Stanford work on situation semantics (CSLI), third on several versions of logic.
  • Slides to 6 - Examples of Dog, Chicken, Mechanic and Car, the sarcasm context, phrases in known contexts and situations understandable by children, even language learned by 3-year old child describing and understanding but not yet interpretable by computers.
  • Slide 7-8 - World, model (Tarski style) and theory (true, false) only some of them are useful for engineering. Neuroscientist Damasio describes – brains form maps and images and in combination yield expressions in language and model theoretic yet flexible semantics.
  • Slide 10-11 – situation related to language and logic, and mapping between them in the context and similar background situations discussed. Determining context from conversation without full sentences or speech describing background is difficult and examples were provided. Purpose for situation important.
  • Slide 12 – Boundary of situation determined not only by physical attributes but also by extending senses (telescopes and microscopes) and psychology, computer methods, agents etc., examples provided in subsequent slides.
  • Slide 16-19 - described situation theory, Stanford and Devlin’s work and an information element Infon, or combination in logic and meta language (Tarski invented word). It is recognized that much more information is needed. Logic and meta language IKL and “that” and ist were described as covered in session on 13 Sept but more in detail. FOL and CL used by Tarski as part of meta-language, conditions of truth related them to domain and multiple hierarchies.
  • Slide 21-22 – Semantics based on IKL to explain issues of multiple situations and propositions, through conceptual graph, nested contexts, can be translated to CLIF and use semantics as in IKL. A situation is a meaningful region of space-time described by the proposition stated by the nested CG. Proposition and logic situations were described by John Sowa through conceptual graphs representing same English sentence. Two notions encapsulating one sentence.
  • Slide 23 - IKL and CL, Pat Hayes work, with “that” operator gives you met-language
  • Slide 24: Pat Hayes and John Sowa agree that IKL is a network logic.
  • John: IKL is much richer language and takes different types of logic. Other strengths of IKL described by John.
  • Slide 27: use of “That” if CLIF and turns it into a name of a proposition uses a theory that sentences that are similar are taken as same proposition, IKL is more general. Likewise slides 28-29 describe other logics used in IKL. Compared n-ary and Greek terms Monad. Modal languages and IKL were described.

Follow on discussion

  • Discussion continued mostly among John Sowa, Douglas R Miles and David Whitten also some inputs from Janet Singer, concentrating on topics related to Microtheories (Mt), Cyc. CycL and IKL. Also discussed were specific use of terms “AND”, “That” and “Ist” in the topic of contexts. It was told that there are 12 different ways of lifting vocabularies etc., using Mt. Doug Lenat confirmed the IKL is sufficient to describe CycL but not without CL and “That” operator. John Said that IKL provided you with properties: you get all the power of CycL, modal logics, context logics, situation theory, etc., with just one operator added to Common Logic: 'that'.

2017-09-27 Third Research Session


  • (to be continued)

Meetings

... further results