Date

Attendees

Discussion items

Focus between now and September

  • Transformation Catalog - Yan - big priority
    • will need feedback on implementation - will deploy a version
  • Galaxy Skill - Sandeep
    • working it on getting it to run as regular extractor on Nebula
  • Green Index - Sandeep
    • removing it from Google Map
    • made simple extractor - showed them how to use clients
    • implement one with ML
  • Deep Learning - Dr. Nam
    • Data from SMU
  • Hathi Trust - Derek/Boris
  • Punch Card - Greg
  • Semantic Annotation v2 - team - biggest priority
    • Hydroshare to Clowder
    • Semantic
    • Unit Conversions
  • Hardening of Clowder
    • update old libraries
    • clean up


TimeItemWhoNotes


Yan

Transformation Catalog

search, detail page, post form, graphQL

Yan demo

Issue - Card tile display

Need to know which fields are required for enter new transformations

Please deploy



Sandeep
  • Met with Matt Browning and Daniel Fleischer and demoed BD CLI - made note to follow up July 1
  • Getting images in batches of 60,000 for a few years
  • Set up folder for student who had interest in BD Galaxy - sent email
  • To spend 4 more hours on the Galaxy skills extractor refactoring and call it done.


Mark
  • Checking into the timeouts. dbpedia cleared itself finally, nlp extractors run without errors and quickly when run on a new file in a personal dataset, not sure why timeouts are being reported here
  • Uploaded a Dockerfile for bdfiddle as a WIP


Deren


Greg

It is going well, but not ready as yet.. I also just got a separate item off my plate for the week. So I hope to report more progress on Monday. It is not the kind of extractor we'd want to run on every image. So I wonder how we make an extractor that only runs on demand.

I thought I remember that it has been done with others and I can ask in #browndog or #clowder



Luigi1.6.1 release for clowder went out


Dukyun Nam

Showed Running Tensor Flow wiki page - I don't have access in NCSA wiki

Showed model for utility poles

Luigi suggested looking at sidewalks - but manual data has utility pole noted in comments to compare - OK

Luigi and Rob will work with Dr. Nam to set up a VM on BD for processing



Rob

Back after vacation and conference

- geotiff needs a lot of work - there are 4 or 5 different geotiff

there is no extractor associated with NCSA Geotiff queue (should be NCSA Geotiff Metadata in the test)

Luigi - there is only preview for tiff, and preview for shape, and a preview that gives metadata for bounding box

Can Yan Zhao - change bdtest to use ncsa.geotiff.metadata for geotiff extractor instead of ncsa.geotiff (lets do this on the dev server)

NLPLanguage - needs a queue created - update tests on production - names are different

To Dos - Tasks


  •