First impressions

• INCEpTION User Guide

  • Docker works:
    • docker run -it --name inception -p8080:8080 ghcr.io/inception-project/inception:35.1
    • $ docker run -it --name inception -v /srv/inception:/export -p8080:8080 ghcr.io/inception-project/inception:35.1
      • data will be in /srv/inception

• Creating a project automatically fills it with sample data:

• Tagsets

  • Export format is:

  {
    "name" : "BBK",
    "description" : null,
    "language" : null,
    "tags" : [ {
      "tag_name" : "aaa_human_processed",
      "tag_description" : null
    }, {
      "tag_name" : "block",
      "tag_description" : null
    } ],
    "create_tag" : false
  }
  

  • Import format: I can get it do do only txt, one tag per line, first line is name of tagset

• A layer has to be linked to a feature (string) which then can be linked to a tagset: (INCEpTION User Guide)

  • then you can add keybindings manually
  • and the “editor type” for the tag list is neat, “Radio group” works nicely for tagsets it doesn’t consider small

• annotations get saved automatically

• in the viewer, you can set dynamic for annotations differing based on color

Export

• Admin->Export can
  • export the entire projects
  • project + separately a copy of the anntations

Resources

• Not-official tutorial: 12. INCEpTION — New Languages for NLP