dkpro-core icon indicating copy to clipboard operation
dkpro-core copied to clipboard
verified publisher icon dkpro

Added Glove.6B files to the dataset resources

Open teaCube opened this issue 8 years ago • 4 comments

Tried to create the necessary yaml-file for downloading the Glove.6B files. Please let me know if it is missing something.

teaCube avatar Mar 22 '17 21:03 teaCube

I notice that before adding this to DKPro Core, probably some additional stuff needs to be considered:

  • Should it really be version 1.0.0 or rather a timestamp?
  • We are not using the original format but a custom binary format, so maybe the ID should be "glove-dl4jw2v"?
  • Maybe better one dataset descriptor per file to allow a more fine-grained download?
  • maybe better also add a timestamp/version number to the filenames on UKP public?
  • is UKP public the place where this data should stay?

reckart avatar Mar 22 '17 21:03 reckart

Answer to data upload repository issue: Figshare can be considered as one option for sharing the dataset https://figshare.com/ They provide a DOI for the landing page of the dataset but I'm not sure if they provide a permanent link for the dataset file.

maxxkia avatar Jun 07 '17 10:06 maxxkia

CLARIN LINDAT or Zenodo would IMHO be more applicable options. Or just downloading the original files from Stanford and doing the conversion locally afterwards.

reckart avatar Jun 07 '17 10:06 reckart

Can one of the admins verify this patch?

ukp-svc-jenkins avatar Apr 30 '19 11:04 ukp-svc-jenkins

No relevant anymore.

reckart avatar Feb 28 '24 19:02 reckart