|Session||Synthesis Session 1|
|Date/Time||27 February 2019 17:00 GMT|
|9:00am PST/12:00pm EST|
|5:00pm GMT/6:00pm CET|
Ontology Summit 2019 Synthesis Session 1
The aim of this week's session is to synthesize the lessons learned so far on the tracks that are under way. Each track has met once, and so we will have gained insights from a combination of invited speakers, chat log comments and blog page discussions.
A second synthesis session will take place after the tracks have met again, and that, along with today's session outcome, will form the basis of this year's Ontology Summit Communiqué.
- Introduction: Ken Baclawski (See summary below)
- The track co-champions will give summaries of their respective tracks:
- Commonsense (See Resources Section below)
- Financial Explanations
- Medical Explanations
- Explainable AI Ram D. Sriram and Ravi Sharma Synthesis
- Video Recording
Summary of Ontology Summit 2019 Sessions
There were 9 sessions so far. Each session had the proceedings (from the chat room) and a recording (one audio recording and the rest video recordings). The following are the speakers with links to their presentation slides (when they were provided) and the recordings.
|11/14||John Sowa||Explanations and help facilities designed for people||Slides||Video|
|11/28||Ram D. Sriram and Ravi Sharma||Introductory Remarks on XAI||Slides||Video|
|Derek Doran||Okay but Really... What is Explainable AI? Notions and Conceptualizations of the Field||Slides|
|12/05||Gary Berg-Cross and Torsten Hahmann||Introduction to Commonsense Knowledge and Reasoning||Slides||Video|
|1/16||Ken Baclawski||Introductory Remarks||Slides||Video|
|Gary Berg-Cross and Torsten Hahmann||Commonsense||Slides|
|Donna Fritzsche and Mark Underwood||Narrative||Slides|
|Mark Underwood and Mike Bennett||Financial Explanation|
|Ram D. Sriram and David Whitten||Medical Explanation|
|Ram D. Sriram and Ravi Sharma||Explainable AI||Slides|
|1/23||Michael Grüninger||Ontologies for the Physical Turing Test||Slides||Video|
|Benjamin Grosof||An Overview of Explanation: Concepts, Uses, and Issues||Slides|
|1/30||Donna Fritzsche||Introduction to Narrative||Audio only|
|Ken Baclawski||Proof as Explanation and Narrative||Slides|
|Mark Underwood||Bag of Verses: Frameworks for Narration from Cognitive Psychology||Slides|
|2/6||Mike Bennett||Financial Explanations Introduction||Slides||Video|
|Mark Underwood||Explanation Use Cases from Regulatory and Service Quality Drivers in Retail Credit Card Finance||Slides|
|Mike Bennett||Financial Industry Explanations||Slides|
|2/13||David Whitten||Introduction to Medical Explanation Systems||Slides|
|Augie Turano||Review and Recommendations from past Experience with Medical Explanation Systems||Slides|
|Ram D. Sriram||XAI for Biomedicine||Slides|
|2/20||William Clancey||Explainable AI Past, Present, and Future–A Scientific Modeling Approach||Slides||Video|
Conference Call Information
- Date: Wednesday, 27-February-2019
- Start Time: 9:00am PST / 12:00pm EST / 6:00pm CET / 5:00pm GMT / 1700 UTC
- ref: World Clock
- Expected Call Duration: 1 hour
- The Video Conference URL is https://zoom.us/j/689971575
- iPhone one-tap :
- US: +16699006833,,689971575# or +16465588665,,689971575#
- Dial(for higher quality, dial a number based on your current location): US: +1 669 900 6833 or +1 646 558 8665
- Meeting ID: 689 971 575
- International numbers available: https://zoom.us/u/Iuuiouo
- iPhone one-tap :
- Chat Room
- Alessandro Oltramari
- Alex Shkotin
- Andrea Westerinen
- Bobbin Teegarden
- Dave Whitten
- Douglas R Miles
- Gary Berg-Cross
- Janet Singer
- John Sowa
- Ken Baclawski
- Mark Fox
- Mark Underwood
- Mike Bennett
- Ram D. Sriram
- Ravi Sharma
- Russell Reinsch
- Spencer Breiner
- Steve Ray
- Terry Longstreth
- Todd Schneider
- Torsten Hahmann
- William Clancey
Here are some ideas for a working synthesis outline for Explanations (Gary Berg-Cross)
1. Meaning of Explanation [An explanation is the answer to the question "Why?" as well the answers to followup questions such as "Where do I go from here?"] – there are range of these
- Grosof deductive Proof , with a formal knowledge representation (KR) – is the gold standard, but there are many types with different representations
- – E.g., natural deduction –HS geometry there is also probabilistic
- Causal model Explanations
There are a range of concepts related to explanation
- Source or provenance, say of a rule
- Transparency in origin
- Ability to explore and drill down
- Focus on the subject on hand
Additional Aspects/Modifiers of explanation:
- Summarization, grain (coarse vs. fine), drill-down, elaboration
- Partial vs. complete
- Approximate vs. precise
- Structuring of inference in presentation
- Assumptions and presumptions
- Targeting to user knowledge and goals (i.e., user model)
- Natural language (NL) generation
- Graphical presentation
- Use of Terminology (source and validity)
- Understand-ability and presentability
Trending-Up concepts of explanation
- Influentiality – , heavily weighted hidden nodes and edges effect
- Reconstruction – simpler / easier-to-comprehend model
- Lateral relevance – interactivity for exploration
- Affordance of Conversational human-computer interaction (HCI)
- Good explanations quickly get into issue of understanding & meaning since much meaning involves background knowledge and commonsense and lies in the implicit and unspoken.
- What does it mean to understand , follow and explain a set of instructions?
2. Problems and issues
- From GOFAI
- An early goal of AI was to teach/program computers with enough factual (often commonsense) knowledge about the world so that they could reason about it in the way people do which is not strictly logical is some circumstances
- early AI demonstrated that the nature and scale of the problem was difficult.
- one reason is that simple, direct approaches like rule based systems were brittle.
- People seemed to need a vast store of everyday, background knowledge for common tasks. A variety of background knowledge was needed to understand & explain decisions
- Do we have a small, common ontology that we mostly all share for representing and reasoning about the physical world?
- it remains challenging to design and evaluate a software system that represents commonsense knowledge and that can support reasoning (such as deduction and explanation) in everyday tasks. (evidence from modified Physical Turing Tests)
- PRAxIS work (Perception, Reasoning, and Action across Intelligent Systems)
3. From XAI
- Performance vs. Explainability: DARPA XAI Program
- While DL performs well is is very low on explainability which decision trees are the reverse
- One context of recent work is that of Deep machine-learning systems. Explanations for their decisions can be problematic since one cay say what they learn is a dimension space of numbers that have no words of any kind, so explanation in natural language is not immediately available.
- One approach to handling this problem is called Deep Explanation which uses modified deep learning techniques to learn explainable features as part of what it learns.
- More traditional GOFAI approaches may develop Interpretable Models using techniques that learn more structured, interpretable, causal models.
- Concept of Humagic Knowledge
- Bridging from sub-symbolic to symbolic -ontologies help constrain options
4. Application areas
- Automated Decision Support for Financial Regulatory/Policy Compliance
- Has requirements like competency Qs it needs to explain
- Examples of successes? Rulelog’s Core includes Restraint bounded rationality
5. Relevance and relation to context
6. Synergies with commonsense reasoning
- Spatial and physical reasoning are good areas.
7. Success stories/systems
- ErgoAI Architecture ?
- Issues Today in the Field of Explanation /Questions
- How do we evaluate these ontologies supporting explanations and commonsense understanding?
- How are these explanations ontologies related to existing upper ontologies?
- In the future, we’ll share meanings with
computers, AIs, and robots. And that makes meanings matter even more. But it remains a hard problem.
- Smart systems may have to be embodied & have
sentience—the capacity to feel, perceive, or experience subjectively.
- Benefits of Explanation (Grosof)
- Semi-automatic decision support
- Might lead to fully-automatic decision making – E.g., in deep deduction about policies and legal – especially the business and medicine topics.
- Useful for Education and training, i.e., e-learning – E.g., Digital Socrates concept by Janine Bloomfield of Coherent Knowledge
- Accountability • Knowledge debugging in KB development
- Trust in systems – Competence and correctness – Ethicality, fairness, and legality
- Supports Human-machine interaction and User engagement (see Sowa also)
- Supports Reuse / and guide choice for transfer of knowledge
9. Contemporary Issues
- Confusion about concepts – Esp. among non-research industry and media – But needs to be addressed first in the research community
- Mission creep, i.e., expansivity of task/aspect – Esp. among researchers. E.g., IJCAI-18 workshop on explainable AI.
- Ignorance of what’s already practical – E.g., in deep policy/legal deduction for decisions: full explanation of extended logic programs, with NL generation and interactive drill-down navigation – E.g., in cognitive search: provenance and focus and lateral relevance, in extended knowledge graphs
- Disconnect between users and investors
- (Ignorance of past relevant work)
- Some envision a fruitful marriage between classic logical approaches (ontologies) with statistical approaches which may lead to context-adaptive systems (stochastic ontologies) that might work similar to the human brain.
[A preliminary summary of the track on CommonSense Knowledge and Reasoning along with some discussion of its Relation to Explanation.]