Usage-based Data Discovery - Episode 2

From Earth Science Information Partners (ESIP)
Revision as of 16:50, March 25, 2020 by Hyoklee (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

Participants

  • Chris Lynnes
  • Joe Lee
  • Irina Gerasimov
  • Adrienne Canino
  • Megan Carter
  • Ed Armstrong
  • Mark Parsons
  • Bob Downs
  • Beth Huffer
  • Jonathan Blythe
  • Doug Newman
  • Steve Olding

Current Actions

  • Done: Lynnes - write up some text for Megan's newsletter
  • Done: Lynnes set up regular meeting with ESIP G2M, some time other than 4th Thursday
  • TBD - spin off a subgroup to work on use cases
  • Done: Lynnes/Newman - generate a sample export of EOSDIS knowledge graph
  • Done: Blythe - provide research_prov_model info on use of W3C Prov model

Agenda

  • Use cases for connecting Data with Usage
    • Best Practice Use Case: Which rainfall dataset is most often used for flood prediction?
    • Impact Use Case: How many climate studies use TRMM 3B42 rainfall?
    • Congruency Use Case: Do field data, airborne lidar, and satellite data match (MAAP)? Do data from sensors on the same satellite match (TerraFusion)?
  • Sample graphs
  • Any other business?
  • Agenda for next telecon…

Notes

  • Use Cases: Participating: Bob, Jonathon, Beth, Ed

Best Practice Use Case

  • Ed: interested in interdisciplinary, datasets used together WITHIN a given application
  • Mark: simple, single dataset use case + complex interdisciplinary use cases
  • Beth: dataset interconnections circles back to data transformations
  • Irina: validation requires MANY different datasets being used together, shows up in separate of research papers that focus on the application vs. those that focus on validation
  • J.B.: discovery / search process: present datasets relevant to their problem, e.g., homeowner types in more colloquial term, interpret what the user's real concern is
  • Adrienne (Axiom Data Science): how do we interact with users where they know what they are getting back; users expect fancy searches, but they need to be transparent (document how they got there) - show the graph!
  • B. Downs: combine with location of interest
  • JB: maybe too big a tech lift at this point?
  • Mark: location is relevant, sure, but how does it
  • Chris: dataset used Gulf of Maine, might be useful in Puget Sound
  • Mark: identify desired outcome

IMPACT Use Case

  • In addition to how many, what type of use cases
  • Beth: Take me to the articles about the data
  • Irina: Data providers want to know about applications of their data; ask them if they know how their data are being used
  • Bob: What other data were used in conjunction with the datasets in a given climate study
  • JB: datasets commonly used together (Mark: closer to Beth's use case variant)
  • JB: IMPACT might be a roll up of lots of datasets
  • Mark: may want to contrast similar instrument, different algorithms, (Chris: or different platforms)
  • Ed: roll up by Societal Benefit Area

Next Agenda

  • Irina, Beth more graphs, how to make them
  • Adrienne: User story, aiming to create at the end
  • Ed: Using graphs for discovery, esp. related datasets, practical examples
  • Mark: Take one of these use cases, work a user story end to end to see what the graph looks like, including the tools
  • Bob: examples of graphs, something about the process we are going through, getting from here to there
  • Megan: Session proposal, ESIP Collaboration Slide

New Actions