Open Dynnammo opened 1 year ago
Would you write your environment? Thank you!
If you upload the inputs through a json based format (JSON/L), if the key in the json doesn't match the "Column Label" text, it will be uploaded as Metadata
Ie.
{"text": "blah", "label": "blah", "your_thing": "blah"}
If you upload the inputs through a json based format (JSON/L), if the key in the json doesn't match the "Column Label" text, it will be uploaded as Metadata
Ie.
{"text": "blah", "label": "blah", "your_thing": "blah"}
Yes, I got it. What I was asking for is precisely if it would be relevant to add your answer in the documentation. Currently (or maybe I read it wrong), the Import dataset section is quite scarce about this .
@Dynnammo We could add in some documentation for it. I'd also like to see the relation labels supported so we could document that as well!
I don't think I have work with what you call relation labels yet, but I'm ready to learn more about it ^^. You can assign me on this issue, I'll try to find some time to work on it by the end of the week :+1:
Hello, @Dynnammo
Great suggestion. I'm looking forward to your work.
@Dynnammo I have an open issue about supporting relation labels that should point you in the right direction. -> https://github.com/doccano/doccano/issues/2039. If there is anything I can help with, please let me know!
:warning: Disclaimer : newbie on this project as a contributor (however I'm a 2-years users of Doccano, you're product is 🔥). This issue is about documentation and I know you recommend making a PR first, but since my time is a bit constrained I first would like to have insights about the need of this :warning:
Context
As an a analyst, I work a lot on text classifications that are extracted from an external platform. These text chunks have each an ID that I must have after annotation, which I could handle thanks to the metadata field when I import data.
However, as I dive through the documentation (https://doccano.github.io/doccano/tutorial/) on the section of importing dataset, I find not trace of such possibility, which can be quite useful.
I would like to add a section in the import a Dataset section that gives insight about how the file can handle metadata, and more widely informations that are shown nearby annotations. Am I suggesting something worth ?
System information