From OntologPSMW

Jump to: navigation, search
[ ]
    (1)
Session Sean Gordon
Duration 1 hour
Date/Time 22 April 2020 16:00 GMT
9:00am PDT/12:00pm EDT
5:00pm BST/6:00pm CEST
Convener KenBaclawski
Track Use Cases

Contents

Knowledge graphs, closely related to ontologies and semantic networks, have emerged in the last few years to be an important semantic technology and research area. As structured representations of semantic knowledge that are stored in a graph, KGs are lightweight versions of semantic networks that scale to massive datasets such as the entire World Wide Web. Industry has devoted a great deal of effort to the development of knowledge graphs, and they are now critical to the functions of intelligent virtual assistants such as Siri and Alexa. Some of the research communities where KGs are relevant are Ontologies, Big Data, Linked Data, Open Knowledge Network, Artificial Intelligence, Deep Learning, and many others.     (2A)

Agenda     (2B)

  • Sean Gordon Prototyping an Open Knowledge Network for Spatial Decision Support Slides Video Recording YouTube Video     (2B1)
    • Abstract: The proliferation of online data and mapping technologies has greatly increased access to and utility of spatial decision support systems (SDSS). SDSS combine data, tools, models, and maps within a user interface to help people make decisions. SDSS are used across a diversity of problem domains, including public health, emergency management, city planning, education, natural resource management, public safety, transportation, utilities, and the delivery of public and private services more generally. Despite many successful applications, spatial decision support contributions are limited by challenges in integrating information across complex organizational networks and across an array of data and tools developed for narrow (often disciplinary) applications. Most SDSS are effectively closed systems: (a) the components are highly coupled and cannot be reordered, (b) users do not have the flexibility to create new desired functions; and (c) users cannot easily combine the tools or data with other systems.

      In September 2019 a group of SDS academics and professionals received a 1-year phase 1 Convergence Accelerator grant from NSF to plan an "Open Knowledge Network for Spatial Decision Support (OKN-SDS)," which would use semantic technologies, along with other participatory and technological methods, to help bridge these organizational, disciplinary and technological boundaries. Phase 1 focus has been on user needs assessment and technology prototyping, using a human-centered design approach. Based on our team's existing work, we engaged ~60 potential users in the areas of wildland fire, water quality, and biodiversity conservation. We synthesized user needs and then brainstormed technology solutions, drawing from existing tools where available and creating new prototypes where needed. Results include semantic tools for finding geospatial datasets, models, and workflows, as well as a prototype knowledge graph for addressing water quality issues in the Puget Sound in Washington state. (And for the more general public, we also have an "explainer video.")     (2B1A)
    • Bio: I'm a Research Assistant Professor with the Institute for Sustainable Solutions at Portland State University in Portland, Oregon. My research focuses on the design and use of information systems for supporting natural resource decisions and the organizational sociology of how these tools fit (or don't) into institutions and governance networks (the science-policy interface). Although I'm far from a "working ontologist," I've participated in a couple of projects which have built web portals using semantic technologies, including the Spatial Decision Support Consortium and the Forest Management Decision Support Systems Community of Practice. I'm now the principal investigator for an NSF-funded project titled "Open Knowledge Network for Spatial Decision Support," which is the topic of my presentation today.     (2B1B)


Conference Call Information     (2C)

Attendees     (2D)

Discussion     (2E)

[12:06] Ravi Sharma: Sean - what is connection between Earth science Geospatial mapping and convergence acceleration NSF?     (2E1)

[12:07] Ravi Sharma: does that mean data to be available to all just metadata or full datasets?     (2E2)

[12:08] Ravi Sharma: OKN - KGs of private companies (notes)     (2E3)

[12:10] Ravi Sharma: Sean - I see sudden switch to social network from earth science datasets?     (2E4)

[12:14] Ravi Sharma: Sean are - there specific use cases where spatial decision support has brought measurable benefits - say in water, fire, etc?     (2E5)

[12:18] Ravi Sharma: Sean - it looks like you are harmonizing vocabularies across domain in the clusterplot and discovering new relationships and it looks like this would lead to ontologies of the multiple domain (entities and Relations)     (2E6)

[12:19] Ravi Sharma: Sean - how much improvement in ontology did your team achieve?     (2E7)

[12:21] Ravi Sharma: Sean - why are there no overlaps in ontologies say between WQ and Questions?     (2E8)

[12:22] Ken Baclawski: The slides are available at http://bit.ly/2KquIjr     (2E9)

[12:23] Ravi Sharma: Sean - why did you not use FDGC Metadata, NASA metadata from DAACs and also from Dublin Core?     (2E10)

[12:25] Ravi Sharma: Sean I know my list does not cover recent Drone or aerial thematic repositories so you will need clever search to add these to NASA NOAA and hydromete and USGS EROS Data and datasets for discovery?     (2E11)

[12:26] Todd Schneider: Sean, was a foundational ontology used in any way (e.g. analysis, categorization)?     (2E12)

[12:29] Todd Schneider: What form do any rules take?     (2E13)

[12:29] Todd Schneider: What reasoner(s) are used?     (2E14)

[12:30] Todd Schneider: Has change management of any ontologies been addressed?     (2E15)

[12:34] Ravi Sharma: Sean -- saw you create repository as well info system but     (2E16)

[12:35] Ravi Sharma: Sean -- when can we see KGs for decision support?     (2E17)

[12:37] Janet Singer: This is an interesting illustration of the argument John S. has been making about real work using microtheories and natural language.     (2E18)

[12:38] Janet Singer: @Todd: The idea that using a foundational ontology necessarily makes a semantic technology more evolvable is debatable     (2E19)

[12:40] Jiaxin Du. NJIT: The BERT and other models are used in our QA capabilities. They can be used in factual QA for sure. And there are lots of work combining knowledge graph with text data. If a question cannot be answered by a knowledge graph, it will go to related webpages to look for answers.     (2E20)

[12:41] Todd Schneider: A foundational ontology can help with 'integration'.     (2E21)

[12:43] David Eddy: Here's a scary thought... if one is to believe Wikipedia, DevOps was "discovered" in 2009.     (2E22)

[12:43] David Eddy: point being... solid exposure to unbypassable change management is very, very, thin.     (2E23)

[12:44] Janet Singer: J Sowa asks what kind of magic tools would the SDSS benefit from?     (2E24)

[12:44] David Eddy: how's this for scary... you're the manager responsible for make this fly in formation; https://xebialabs.com/periodic-table-of-devops-tools/     (2E25)

[12:48] Jiaxin Du. NJIT: Reference of technology we use: Paper: Xiong, Wenhan et al. Improving Question Answering over Incomplete KBs with Knowledge-Aware Reader. Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics (2019): n. pag. Crossref. Web.https://github.com/dujiaxin/Knowledge-Aware-Reader  ; Paper: Ding, Ming et al. Cognitive Graph for Multi-Hop Reading Comprehension at Scale. Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics (2019): n. pag. Crossref. Web. https://github.com/dujiaxin/CogQA     (2E26)

[12:49] Janet Singer: @David: but putting together a periodic table cartoon is the magic part that makes a field into a science!     (2E27)

[12:50] John Sowa: For references, Google FCA Formal Concept Analysis     (2E28)

[12:51] Yuanyuan Tian: PolarHub link: http://cici.lab.asu.edu/polarhub3/     (2E29)

[12:52] John Sowa: Uta Priss has put together many of the FCA resources: https://www.upriss.org.uk/     (2E30)

[12:52] David Eddy: @Janet... and in a vastly slower & smaller time... it took 200 years to evolve Isaac Newton's alchemy (bleeding edge science) to Mendeleev's Periodic Table.     (2E31)

[12:54] John Sowa: For demos based on WordNet and Roget's Thesaurus, see://www.ketlab.org.uk/     (2E32)

[12:55] John Sowa: The FCA tools are free and open source. You can define a "context" as a set of words, each specified by a list of features.     (2E33)

[12:56] John Sowa: Then you feed that "context" into FCA, push a button, and you get a complete and consistent lattice of all the terms.     (2E34)

[12:58] John Sowa: Your context can contain a vocabulary of thousands of words, and you can add multiple, independently developed contexts (vocabularies) to the mix     (2E35)

[12:59] Bobbin Teegarden: @JohnS is the 'context you feed FCA tools a list or graph? is the result a hierarchical 'taxonomy' or a real graph shape?     (2E36)

[12:59] John Sowa: When you push the button, all the words from all the sources automagically appear in a lattice that is guaranteed to be consistent with the data you entered     (2E37)

[13:00] John Sowa: Of course, the lattice is only as good as the data you entered.     (2E38)

[13:01] Bobbin Teegarden: @JohnS is the lattice DAG shaped?     (2E39)

[13:01] John Sowa: But after you get the lattice generated, you can use the FCA tools to browse that lattice and look for regions that may require the humans to redefine some of the terms.     (2E40)

[13:02] Ravi Sharma: great talk Sean and Team     (2E41)

[13:03] John Sowa: A lattice is a directed acyclic graph in which every pair of nodes have a minimal common supertype and a minimal common subtype.     (2E42)

[13:04] Janet Singer: @David: And Mendeleev's is only one scheme https://en.m.wikipedia.org/wiki/Alternative_periodic_tables     (2E43)

[13:04] John Sowa: If two terms have nothing in common, their minimal supertype is the universal type at the top.     (2E44)

[13:05] Ravi Sharma: It is a Tragedy that the white house decision support network and engines were not used to fight supply shortages in coronavirus health needs     (2E45)

[13:05] John Sowa: If they are inconsistent, their only common subtype is the absurd type at the bottom.     (2E46)

[13:06] Bobbin Teegarden: @John thank you for that!     (2E47)

[13:06] Todd Schneider: Meeting ends @13:06 EDT.     (2E48)

[13:06] John Sowa: Another way to think of a lattice is that it is a tree at both ends.     (2E49)

[13:07] John Sowa: Ravi, the tragedy is that the guy in charge is a moron.     (2E50)

Resources     (2F)

Previous Meetings     (2G)


Next Meetings     (2H)