doccano - 机器学习从业者的开源文本注释工具

Enhancements, Bug Fixes, and Other Changes

Enhancements

Update packages

• #821 Update Django
• #823 Update websocket-extensions
• #898 Update lodash
• #909 Replace django-pyodbc-azure with django-mssql-backend
• #677 Fix mkdocs and mkdocs-material version

Improve the documentation

• #729 Update FAQ to include how to logout
• #779 Improve installation documentation
• #742 Update rebuild FAQ
• #832 FAQ correction
• #878 Rewrite FAQ about changing port number
• #912 Add link to the documentation
• #684 Clean document structure
• #780 Update project_structure.md
• #781 Adding annotators documentation improvement

Improve annotation

• #489 Single class classification
• #495 Speech to text transcription
• #817 Display newlines for sequence-labeling projects in old frontend
• #908 Keep the new line in the text classification
• #818 Show thumbs up icon for doc approved status on the left side view
• #766 Change of permission for annotation edits

Improve a project page

• #678 Updated statistics page user data graph
• #671 Clarified input format
• #872 Feature/excel conll import
• #899 Highlight current page in the side bar
• #911 Update error message on login failure
• #838 Improve home page text
• #904 Fix missing icon

Improve setup and middleware

• #301 Added middleware that enables automatic login based on HTTP headers
• #795 Use Alpine as root image
• #791 Remove legacy endpoint
• #812 Dockerfile cleanup
• #801 Add simple /v1/health endpoint
• #774 Added flag for allowing downgrades while using apt
• #673 Remove middleware duplication
• #843 Turn off caching on apk and pip
• #787 Avoid username is taken
• #862 Make google-analytics configurable

Bug fixes

• #841 Fix aws deploy
• #855 Fix incorrect current user name
• #885 Fix choosing filter twice results in 404 error
• #897 Fix pagination and transition issues
• #903 Fix 400 error on text classification
• #863 Fix error on Django Admin search
• #864 Fix problem to reset label color to #FF0000
• #865 Fix unworked cancel button in label page
• #900 Enable to close a dialog
• #916 Allow annotator to update/delete other's annotations
• #918 Fix ApproveLabelsAPI not to show other's annotations if it's checked
• #680 Added username in localstorage
• #693 Add CORS whitelist

Bug Fixes, Enhancements, and Other Changes

Bug fixes

• #513 Enable to display non-ASCII metadata
• #518 Enable to redirect to root page
• #587 Fix the annotation toolbar vanishment
• #530 Fix filter initial state
• #545, #613 Fix login/logout problem

Enhancements

• #604 Enable to show the current user name

Bug Fixes, Enhancements, and Other Changes

Bug fixes

• #654 Enable to handle multiline document in sequence labeling
• #640 Enable to handle real enter event
• #611 Fix broken links in tutorial.md
• #607 Fix duplicate key error
• #532 Fix flags set on project creation are discarded
• #539 Fix build failure
• #527 Fix collaborative annotation for stats and filter

Enhancements

• #647 Add API document generator
• #558 Add a docker-compose file for development
• #569 Add project structure documentation
• #474 Avoid redirects with custom port dropped
• #549 Disable left/right shortcut key in input and textarea
• #480 Enable to move any page
• #651 Enable to show error messages on importing labels
• #606 Enable to show menu by default
• #603 Enable to use of digits as shortcut keys
• #535 Enable to import of CSVs with arbitrary column order
• #633 Update nginx.conf to enhance security
• #504 Verification emails on Azure

Other changes

• #602 Update package.json
• #582 Update repository name
• #580 Update libcurl3 version
• #567 Update webpack version

Bug Fixes and Other Changes

• #441 Add volume for SQLite DB in doccano container
• #473 Fix error on first annotation page open
• #470 Fix error on creating a label with an empty shortcut key
• #456 Fix error on shortcut key in text classification
• #465 Remove unused dependencies
• #422 Supports MySQL

Bug Fixes and Other Changes

• Adds X-Frame-Options header to prevent clickjacking.
• Adds DEBUG=False to docker-compose.prod.yml.
• Updates Docker Compose file version to 3.7.
• Updates README.md for v1.0.0
• #350 Adds settings required for SSL.
• #373 Adds linter for Dockerfile.
• #388 Adds GCP Run option as one-click deployment.
• #418 Adds label nonfield error messages.
• #443 Adds label import/export features.
• #363 Configure DJANGO_SETTINGS_MODULE properly.
• #319 Enable sorting of documents in frontend.
• #398 Enable superuser to delete user annotations.
• #390 Improve annotation creation performance.
• #448 Updates tornado dependency on 5.1.1 to not be affected by DDOS.
• #450 Updates project creation form for v1.0.0.

Major Features and Improvements

Doccano 1.0.0 focuses on simplicity and supporting mobile devices, featuring updates like:

• Supporting mobile devices.
• Easy deployment in a production environment on any platform.
• Code simplification to reduce hacktivation energy.

Highlights

• doccano 1.0.0 uses Vuetify as a Vue.js UI component framework. Vuetify provides a variety of material components such as button, card, menu, table, for immediate implementation and intuitive code.
• Easy deployment to a production environment: doccano 1.0.0 users will be able to use the docker-compose.prod.yml to deploy the application. It is composed of several containers such as Web(Nginx), application(Gunicorn + Django), database management system(PostgreSQL). The architecture improves stability, reliability, performance of doccano.
• Frontend codes clean-up, included removing large components, instead, create many small components and use them to create a page.
• Dark mode support: doccano 1.0.0 users will be able to use the dark mode. Generally speaking, it takes quiet time to annotate dataset. As a result, our eyes are tired. By using dark mode, our eyes will be less tired. Friendly to eyes!
  
• Mobile device support: Supporting mobile devices is important because a lot of people in the world don't have their own personal computer but probably they have a smartphone. By supporting mobile devices and those people, it will become easier to utilize crowdsourcing, as a result, we will be able to improve our workflow.