HumanSignal / HumanSignal/label-studio
Export problem of labelled data
- Dominant language
- TypeScript
- Stars
- 28.3k
- Forks
- 3.7k
- Avg merge
- 14h
- Merged PRs (30d)
- 15
Description
Hi There, I need help with the labelled data. I labelled 20 historical cookbooks in English. The only label is "Recipe", so it is not just words but I labelled as "Recipe" entire chunk of text (each recipes separately). Now I would like to use these label data to train a model and use it to extract recipes from the other 3000+ old cookbooks I have. When I download/export either the CONLL2003 or Json, the file don't contain the text I labelled, only number on start and end (I guess is the number of the word). Here I paste part of the json I exported:
[{"id":1,"annotations":[{"id":9,"completed_by":1,"result":[{"value":{"start":204443,"end":205641,"labels":["Recipe"]},"id":"gew3t9fgkg","from_name":"label","to_name":"text","type":"labels","origin":"manual"},{"value":{"start":205642,"end":206204,"labels":["Recipe"]},"id":"0gpIsb7s27","from_name":"label","to_name":"text","type":"labels","origin":"manual"},{"value":{"start":206223,"end":207365,"labels":["Recipe"]},"id":"dPrmlsIT9D","from_name":"label","to_name":"text","type":"labels","origin":"manual"},{"value":{"start":207366,"end":208028,"labels":["Recipe"]},"id":"SQF7Wv8hrZ","from_name":"label","to_name":"text","type":"labels","origin":"manual"},{"value":{"start":208029,"end":208761,"labels":["Recipe"]},"id":"98g-cOnado","from_name":"label","to_name":"text","type":"labels","origin":"manual"},{"value":{"start":208762,"end":209305,"labels":["Recipe"]},"id":"1bsWb_f18v","from_name":"label","to_name":"text","type":"labels","origin":"manual"},{"value":{"start":209306,"end":210246,"labels":["Recipe"]},"id":"2n8--rKvw9","from_name":"label","to_name":"text","type":"labels","origin":"manual"},{"value":{"start":210247,"end":210814,"labels":["Recipe"]},"id":"JiG4AfX_xb","from_name":"label","to_name":"text","type":"labels","origin":"manual"},{"value":{"start":210824,"end":211257,"labels":["Recipe"]},"id":"pZppVzewBC","from_name":"label","to_name":"text","type":"labels","origin":"manual"},{"value":{"start":211258,"end":211384,"labels":["Recipe"]},"id":"-kt9q2Th2N","from_name":"label","to_name":"text","type":"labels","origin":"manual"},{"value":{"start":211428,"end":211651,"labels":["Recipe"]},"id":"OOEUK5Yh-n","from_name":"label","to_name":"text","type":"labels","origin":"manual"},{"value":{"start":211653,"end":212748,"labels":["Recipe"]},"id":"Lvyyy-_5SL","from_name":"label","to_name":"text","type":"labels","origin":"manual"},{"value":{"start":214017,"end":214704,"labels":
....
It keeps going.
As you can see the text is not there, not sure why. Is the Json suppose to look like this?
I must mention that in order to import the txt file that I had stored in my Dropbox folder I used this code on Label-studio:
As you can see in order to import the txt files from my dropbox, for some reason, I had to change the valueType in "url". After that I imported the file as a "Time Series of Whole Text". I also tried to just upld a txt file from my desktop and I had to follow the same procedure.
Not sure if this impacted my Json export at the end.
Am I missing something? How can I use this Json to train a model?
Any help would be greatly appreciated.
Marco
Contributor guide
Assessment
This issue has not been assessed yet.