From OntologPSMW

Jump to: navigation, search
[ ]
    (1)
Session Topic Discussion
Duration 1 hour60 minute
3,600 second
0.0417 day
Date/Time July 24 2019 16:00 GMT
9:00am PDT/12:00pm EDT
5:00pm BST/6:00pm CEST
Convener Ken Baclawski

Contents

Agenda     (2A)

Open-ended discussion of the topics for summer meetings and the next summit.     (2A1)

Property Graphs, Graph Databases, Knowledge Graphs and Ontologies     (2A3)

Conference Call Information     (2B)

Participants     (2C)

Proceedings     (2D)

[11:59] Gary: We might have 2 presentations as background to knowledge and property graphs. Todd and Gary     (2D1)

[12:04] Ken Baclawski: The slides are available at http://bit.ly/2JMlq1H and http://bit.ly/2JNgHg5     (2D2)

[12:12] janet singer: This piece might provide focus for a KR-Ontology-standards topic https://www.linkedin.com/pulse/beyond-low-code-hype-knowledge-graph-driven-alan-morrison/     (2D3)

[12:14] Ken Baclawski: Link to the UCSF conference: http://bakarinstitute.ucsf.edu/open-knowledge-network/     (2D4)

[12:16] MikeBennett: KG is widely used in finance and elsewhere in industry to refer to an e.g. RDF store with Individuals (instances; digital twins) of concepts in an e.g. OWL ontology.     (2D5)

[12:17] janet singer: A strong claim from the above link: Keep in mind that scalable integration and even interoperability of heterogeneous sources can be the immediate payoff of a standards-based knowledge graph     (2D6)

[12:23] Ram D. Sriram: @janet. These KGs should also be able to handle mathematical relationships that arise in real world situations, e.g., mechanics.     (2D7)

[12:25] Ram D. Sriram: @janet. I agree with your comment. My comment is in addition to what you said.     (2D8)

[12:27] ToddSchneider: A couple points to consider in the discussions of 'knowledge graphs': 0) logical consistency, 1) Ontological consistency, 2) Querying/searching.     (2D9)

[12:28] John Sowa1: The term 'knowledge graph ' is a meaningless buzz word.     (2D10)

[12:30] Ram D. Sriram: @John. Good point. We need to first come out with the specifications and discuss how various representations meet these specifications     (2D11)

[12:31] MikeBennett: We can surely agree that an RDF graph with individuals is an exemplar of KG. The question is what else to also call a KG based on generalized observations of features similar to those in RDF.     (2D12)

[12:36] ToddSchneider: The term 'knowledge graph' is a buzz word, but not 'meaningless'. Companies are using it to sell products and organizations are implementing graph databases (i.e., information persistence mechanisms) and labeling them 'knowledge graph'.     (2D13)

[12:37] MikeBennett: The label 'Knowledge Graph' may range over an indistinct semantic space but we can provide some clarity by narrowing it down to the kinds of graph for which there is some ontology.     (2D14)

[12:37] ToddSchneider: I suggested 'knowledge graph' as the subject of the Ontology Summit with the intent of attracting more participants.     (2D15)

[12:39] MikeBennett: Not sure it's always the case that something called a KG in the Semantic Web sense, necessarily has provenance and temporality information. It may be if is e.g. a quad store rather than just RDF triples.     (2D16)

[12:40] RaviSharma: Ravi - sorry for joining late     (2D17)

[12:43] MikeBennett: To the extent this is not new, john's thousand other things are also knowledge graphs then. The difference is, this is how a business understands it can use it to get some business value     (2D18)

[12:44] MikeBennett: It it doesn't have an ontology we should not call it a KG.     (2D19)

[12:51] MikeBennett: KG is a friendlier face for what we have already been doing in SemWeb, making it more attractive for people.     (2D20)

[12:51] MikeBennett: We can describe kinds and levels of KG in terms of what kinds of logic are used to define the semantics of the instances in the data.     (2D21)

[12:52] MikeBennett: We can show some leadership in terms of the principles and best practices in KGs.     (2D22)

[12:53] MikeBennett: If you were to add instance data to a Conceptual Graph, you would have another kind of Knowledge Graph.     (2D23)

[12:54] RaviSharma: from ORiley book: What Are Graphs?     (2D24)

Graphs have a history dating back to 1736, when Leonhard Euler solved the Seven Bridges of Konigsberg problem. The problem asked whether it was possible to visit all four areas of a city connected by seven bridges, while only crossing each bridge once. It wasn't possible.     (2D25)

With the insight that only the connections themselves were relevant, Euler set the groundwork for graph theory and its mathematics     (2D26)

[12:54] MikeBennett: Recognize KG as a business term for something that is marketable (what JFS calls hype).     (2D27)

[12:55] MikeBennett: Suggested title: Hype v Reality? Knowledge Graphs and Applications     (2D28)

[12:55] Ram D. Sriram: Mills Davis posed a question on Ontology Summit discussion     (2D29)

[12:55] RaviSharma: Apache Spark and Neo4J     (2D30)

[12:58] janet singer: I understand Johns concern that switching labels leads to waste of time rediscovering all the same issues over and over again     (2D31)

[12:58] RaviSharma: The book by Amy Hodler and Needham     (2D32)

[13:00] RaviSharma: John what is different in Graph algorithms in this book and Knowledge Graphs     (2D33)

[13:03] janet singer: From Paola Di Maio: However, I would like to say (please note) that knowledge graphs (and related artefacts) are not adequate representation mechanisms to satisfy the KR requirements of systems as being deployed (they are useful but not sufficient) and quite the contrary, contribute to reinforce existing challenges and maybe create more. Much of what is being discus is a waste of time and totally misses the point     (2D34)

I suspect, it may be deliberately designed to deviate community efforts and resources away from the real issues (as always)     (2D35)

[13:07] janet singer: John: if Knowledge Graphs are limited to RDF you cant express OR, NOT, IF-THEN     (2D36)

[13:08] ToddSchneider: Have to go. Have we decided on a/the subject for the next Ontology Summit?     (2D37)

[13:09] MikeBennett: But why would they be so limited? RDF graphs with instances were presented as an example of a /kind/ of KG, not a constraints on what they can be.     (2D38)

[13:10] MikeBennett: What does 'Knowledge Graph' mean? It means that people in industry are finally paying attention to all things we as a community have been developing all along. Let's not shout at them just because we don't like their words. As ontologists we should be better than that.     (2D39)

[13:11] RaviSharma: Improve ML predictions using Graph Algorithms this was a Neo4J webinar recorded     (2D40)

[13:12] janet singer: Agree with Ram to find a positive framing, like : Lessons from Ontologies for Knowledge Graph development     (2D41)

[13:13] MikeBennett: It would be a good idea to reference DOL and characterize kinds of KG based on kinds of logic that are available.     (2D42)

[13:15] MikeBennett: @Janet +1     (2D43)

[13:18] John Sowa1: Title: knowledge graphs, semantic web, ontology. Hype vs. reality.     (2D44)

[13:20] MikeBennett: @Ravi what you missed: there is a hierarchy: graph > knowledge graph. So KG is a kind of G.     (2D45)

[13:20] MikeBennett: @Ravi take a look at the slides.     (2D46)

[13:21] MikeBennett: Also we established that Knowledge Graph > RDF based KG.     (2D47)

[13:22] MikeBennett: This has kind of become a mini-series. Good idea.     (2D48)

[13:22] RaviSharma: thanks Mike I will and summarize my still open Qs     (2D49)

[13:24] RaviSharma: Ram and Janet John has good idea for title     (2D50)

[13:25] MikeBennett: Hype, reality ans misapplication (not for the title but it is a thing: SemWeb assumptions that every ontology has to be for puzzle solving applications)     (2D51)

[13:26] RaviSharma: mike- implies Hype, reality and misapplications?     (2D52)

[13:27] MikeBennett: So in that sense SemWeb (and therefore KG which is mostly the same thing), is overhyped in the sense that it is valuable for one kind of niche application but is over-sold into other applications areas it is not so good for (integration, AI/NLP etc.)     (2D53)

[13:29] MikeBennett: One suggestion was a Summit on standards, a series of KG where the outcome is input to the Summit in terms of potential standards for understanding (DOL) and applying (RDF, CG etc.) ontologies for Knowledge Graphs. As a track?     (2D54)

[13:30] RaviSharma: speaker lead time is longer     (2D55)

[13:30] John Sowa1: I have to sign off now, bye.     (2D56)

[13:31] RaviSharma: Ken - 3-4 speakers in mini series per week     (2D57)

Ken Baclawski: @Ravi: I meant 1 speaker per week, with 3-4 speakers in all at the series to be held in September 4, 11, 18 and 25.     (2D58)

[13:31] Bruce: Thanks for the session, much appreciated, got to go -- bye.     (2D59)

[13:35] RaviSharma: next week same time to be continued.     (2D61)

[13:35] RaviSharma: Ram that was good summary on youtube     (2D62)

Previous Meetings     (2E)


Next Meetings     (2F)