From OntologPSMW

Jump to: navigation, search
[ ]
    (1)
Session Communiqué Discussion
Duration 1 hour
Date/Time November 04 2020 17:00 GMT
9:00am PDT/12:00pm EDT
5:00pm GMT/6:00pm CET
Convener Ken Baclawski

Contents

Agenda     (2A)

  • Continue discussion of the Communiqué for the Ontology Summit 2020: Knowledge Graphs     (2A1)
    • The third draft is now complete. Except for typos and style requirements of the journal, this should be the final draft.     (2A1A)

Conference Call Information     (2B)

Participants     (2C)

Discussion     (2D)

[12:10] ToddSchneider: What about KG Architecture instead of KG Infrastructure?     (2D1)

[12:23] gary: "For the development of the knowledge graph infrastructure in general and the social science knowledge graph in particular, methods of information extraction, entity interlinking, coreference resolution and data fusion are being investigated and applied." from https://www.gesis.org/en/research/applied-computer-science/knowledge-graph-infrastructure/     (2D2)

[12:24] David Eddy: @Todd... total agreement... "architecture" is essentially a meaningless (mean everything) term like metadata     (2D3)

[12:25] gary: " Graph Storage     (2D4)

The first thing we built for the knowledge graph infrastructure is a graph storage module. We adopted an in-house relational data store as the underlying database, on top of which we implemented a node store and edge store such that one can directly perform CRUD " from https://medium.com/airbnb-engineering/contextualizing-airbnb-by-building-knowledge-graph-b7077e268d5a     (2D5)

[12:31] ToddSchneider: Gary, The reference to 'infrastructure' in "...knowledge graph infrastructure is a graph storage module ..." would appear to be software.     (2D6)

[12:35] ToddSchneider: Schematic =def. (of a diagram or other representation) symbolic and simplified.     (2D7)

[12:37] DrRaviSharma: Todd - great thanks, Ram we can accept it     (2D8)

[12:38] gary: @Todd, some infrastructure is SW, but not all. The choice of a graph language is a slightly different example of a technical design decision.     (2D9)

[12:40] ToddSchneider: infrastructure =def. the basic physical and organizational structures and facilities (e.g., buildings, roads, and power supplies) needed for the operation of a society or enterprise.     (2D10)

[12:42] gary: A practical point is that KG infrastructures must be compatible with some existing technologies or people won't adopt them. KGs are a bit easier than very formal ontologies, a point of why they are popular.     (2D11)

[12:54] ToddSchneider: Gary, yes. It usually comes down to cost.     (2D12)

[12:59] ToddSchneider:     (2D13)

This summit has demonstrated that interest in and use of Knowledge graphs continues to grow. Continues to grow in spite of, or perhaps because, there is no single definition. One of reasons for the growth is the changing needs for uses of information. Those uses require more flexibility, extensibility, and dynamism (which Knowledge Graph Systems provide). But more importantly, given the increasing volume of data and attendant uses, explicit semantics is critical to these new and algorithmic intensive uses.     (2D14)

But this summit has also shown that there does not appear to be a focus on creating well crafted schema for Knowledge Graphs where the semantics of the entities represented are made explicit. There is a reliance on, or an expectation of, the semantics or interpretations of the entities represented to be 'carried' by the use of natural language terms or phrases used as their labels. Of course this is not true in all cases. And the reliance on an expectation of a correct understanding or interpretation of natural language terms and phases is not confined to Knowledge Graphs or even ontologies. (Comment by Gary, this means that those KGs would not support reasoning. But we make reasoning ability a feature in our definition of KGs.) Use of a well crafted schema (aka ontology) where the semantics and interpretations are made sufficiently explicit will increase the usability and life expectancy of a Knowledge Graph.     (2D15)

In many situations it may not be necessary to have a well crafted ontology as the schema for a Knowledge Graph. If the use of the Knowledge Graph is very focused (e.g. a narrow application) and is not expected to be in use for a long period of time, then the need for a diligent development is small. However, for information systems that are expected to be in operation for many years or decades (e.g. large organizations or governmental use), then the costs of continually 'corrections' will be greater than the costs of a diligent development.     (2D16)

Though the development of a well crafted ontology (as the schema for a Knowledge Graph) may not be called for, the use of the techniques for creating well crafted ontologies would result in better and more robust Knowledge Graphs. Such techniques include an understanding of Roles, Qualities and the various analysis perspectives - Dependency, Identity, Unity/Plurality, Mereology.     (2D17)

(Gary, one comment. A point to be made here is that some KG work ignores the lessons learned in the history of limitations and problems laid out in Section 2 on historical perspective. This is John's point and worth mentioning or making again.)     (2D18)

[13:07] DrRaviSharma: mike we did not get any inputs, please text if any.     (2D19)

[13:08] DrRaviSharma: David Whitten likewise     (2D20)

Resources     (2E)

Previous Meetings     (2F)


Next Meetings     (2G)