udos 0.0.2

Universal Dataset Open Standard

udos

Universal Dataset Open Standard

A place to develop an open standard for dataset search, acquisition, reusable preparation, and sharing.

Target user stories

• Search outside data (use several sources such as kaggle, github, named urls, etc?)
• Get local copy of data
  • Set rules for automatic updates of data
• Search locally using:
  • General query expansion (using word2vec?)
  • Custom aliases
  • Fuzzy matching
• Construct data feeds by combining multiple data feeds
• Visualize relationships (to understand the data, spot errors, etc.)
• Operable quick models (for quick-and-dirty inference)