New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
New project type: Question and answer #304
base: master
Are you sure you want to change the base?
Conversation
…ed on these files
… question_and_answer
…, it doesn't work yet. we simply put the barebone.
…e ability to have questions associated to a text. This way, we can simply think annotation as a two step task, entirely separated: - create a question => sequence to sequence [already there, but has some issues to be resolved] - answer a question => question and answer [which i am coding at the moment]
… question_and_answer
… is covered as well. It needs as well to bring extra_text as an optional field.
"components//question_and_answer.vue 27:6 Invalid attribute separator found"
… question_and_answer
…tatistics tab: we had problem, and now they are gone :)
…on't need it, so I simply commented it out
…a rea email, we will have to do set up the parameter EMAIL_HOST etc..
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I did a preliminary pass-through of the code without running the branch just to get a conversation started about the pull request. I'm a bit concerned that this introduces quite a lot of new concepts when really the Q&A task is a combination of sequence labeling and document classification. I'm wondering if it would be possible to re-use some of the existing frontend functionality and remix the sequence labeling Vue code with the document classification Vue code. Perhaps via a mixin or something like that?
… question_and_answer
Don't know exactly why... but the new database |
|
still having the issue, any other clue ? |
@guillim The current build failure no longer is related to the environment but instead is a model error:
I'd try to take a look at the schema created by |
…rror detector suggested so
the TextField cannot be used because mssql won't allow it: we need to use CharField instead https://stackoverflow.com/questions/2863993/is-of-a-type-that-is-invalid-for-use-as-a-key-column-in-an-index
…e index 'api_qandaannotation_document_id_user_id_response_start_offset_cfb14664_uniq' exceeds the maximum length of 1700 bytes for nonclustered indexes" So we change from a response of maximum 1000 character to a response of 500 character. Note: this could be problematic in case someone wants su submit a whole text as an aswer in case the text is larger than 500 charcaters for instance
Ok. I finally found out what was the issue with Microsoft SQL Server: they don't accept TextField when used as a unique_together property So I changed it to a CharField. However, it can't be larger than 500 characters, which may be a limitation in case an answer to a question is larger than 500 characters. I guess if this happens, an issue will be raised ! Thanks for the debugging tip @c-w |
Summary:
This pull request has a simple objective: creating a new annotation mode, for the task "question and answer"
Ref: #81
What changed
Backend: 2 models were created (QandAProject & QandAAnnotation ) and all the following related (serializer)
The model Document was slightly modified, adding the text field called extra_text - allows some tasks to store additional information, like a question for instance.
Migrations were created following the model changes as well
Questionable: the necessity of having QandAStorage. At the moment, I guess I could use Seq2seqStorage instead.
Frontend: a new front was added using the component download_qanda.vue. Visual examples for uploading / downloading data were created as well.
Other Info
This is a Draft pull request at the moment. No tests were created yet, and it isn't merged with the latest master branch due to a problem with pyodbc on my side