09:16:52 From Gary Berg-Cross : We should not forget the importance for humans of Abductive Reasoning and the use of mental models.
09:17:30 From Kirstie Bellman : Gary yes! and you should discuss more now
09:18:23 From Gary Berg-Cross : Abduction reasoning involves a major premise/hypothesis is evident/comes to mind but the minor premise and therefore the conclusion only probable.” Basically, it involves forming a conclusion from the information that is known. A familiar example of abduction is a detective’s identification of a criminal by piecing together evidence at a crime scene.
09:32:07 From Janelle Mason : We simulate this with multi-agent systems also, correct?
09:43:02 From Christopher A. Landauer : von Uexküll
09:47:23 From Janelle Mason : Thanks!
10:10:45 From Christopher A. Landauer : Emilio, Lipton, Perlis – Social Processes and Proofs …., Comm. ACM 22 May 1979
10:12:56 From Gary Berg-Cross : From my notes on explanation for our 2019 Ontology Communique: 1 Meaning of Explanation – there are range of these
2 Grosof deductive Proof , with a formal knowledge representation (KR) – is the gold standard, but there are many types with different representations
3 E.g., natural deduction –HS geometry there is also probabilistic
4 Causal model Explanations
5 There are a range of concepts related to explanation
6 Source or provenance, say of a rule
7 Transparency in origin
8 Ability to explore and drill down
9 Focus on the subject on hand
10 Understand-ability and presentability
11 Trending-Up concepts of explanation
12 Influentiality – , heavily weighted hidden nodes and edges effect
13 Reconstruction – simpler / easier-to-comprehend model
14 Lateral relevance – interactivity for exploration
15 Affordance of Conversational human-computer interaction (HCI)
16 Good explanations quickly get into issue of understanding
10:19:46 From Gary Berg-Cross : Issues in Incrementally adding better
Semantics to Knowledge Graphs
Gary Berg-Cross ( gbergcross@gmail.com )
Abstract
Knowledge graphs (KGs) employ a wide range of semantic resources. However, as is true of complex
information systems, harmonizing rich semantic resources requires effort and involves trade offs.
There are practical reasons to start with modest semantics, and then incrementally add enhanced
semantic improvements. For this process there are a number of active research projects that are
developing light, incremental approaches, methods and tools to support an expanding semantic KG
space that has addressed semantic alignment and harmonization. These projects include methods for
using existing semantic relations and entities harmonized across controlled vocabularies, glossaries of
definitions and ontologies. This article discusses examples of incrementally improving the semantics
of less formal schemas that over time is helping to semantically unify richly interconnected
heterogeneous data ……
10:24:52 From Giuseppe D’Aniello : Kirstie regarding solving problems at different levels of abstraction, you may give a look at the Granular Computing paradigm, I’m working on this
10:27:21 From Kirstie Bellman : Wow Giuseppe, let’s add it to our list of growing collaborations!!!
10:27:42 From Ken Baclawski : https://ontologforum.org/index.php/OntologySummit/Publications
10:29:13 From Melita Hadzagic : Chris, would you please write the name of the author of the paper you’ve just mentioned.. on sonar data? Thanks!
10:29:57 From Christopher A. Landauer : Terry Sejnowski – Neural Networks Vol 1 (1988)
10:30:06 From Melita Hadzagic : Great, thank you!
10:30:52 From Andrew Dougherty : https://sissy.telecom-paristech.fr/
11:56:38 From Kirstie Bellman : How do you know when an ontology is wrong?
11:58:33 From Gary Berg-Cross : Ontologies are models so they are approximate but some are useful. They need to satisfy human criteria of making sense to experts but also the proof from accommodating data to provide answers that can be tested for faithfulness.