Keeping It Simple is Hard
Work in progress !
The main goal of this web application is to facilitate and improve the quality of manual annotation of PDF documents (the current scholar publishing standard), by synchronizing PDF and structured extracted content. Annotation of PDF can go beyond bounding boxes or text selection directly on the PDF text layer, which are both imprecise and subject to multiple layout and text quality issues.
The technical approach ofthis application is kept volontary relatively simple, without any database to install/run. The application is a Python FastAPI backend service combined with an Angular web application. The tool uses an embedded SQLite database. Every back-end actions (including authentication) is realized via a REST API. Angular services connect the web components to this API.
Back-end: Python 3.7+, FastAPI, fastapi-users (for secure authentication), SQLITE, sqlalchemy
Front-end: Angular 16.2, PDF.js, Bootstrap 5, recogito-js (and still some remaining JQuery meant to disappear)
Secure authentication requires to indicate SMTP settings (for email after password reset) and OAuth2 log-in requires client credentials depending on the OAuth service where to redirect the user. Data import/export format is JSON only. Annotation guidelines should be written in markdown and put under resources/data/markdown/.
The Bootstrap front-end components are derived from the Sleek theme (MIT license).
- Manage list of datasets and annotation tasks:
- Manage multi-stage user annotation tasks (with reconciliation phase):
- Support direct PDF annotations, as well as excerpt classification and annotations:
Guidelines can be associated to tasks and available during
PDF and excerpts can be pre-annotated.
-
Support user authentication, annotator/curator/admin roles
-
Inter-Annotator Agreement for datasets are calculated automatically based on redundant tasks:
Import and export of dataset and annotations are currently realized with python command lines using JSON (more to come).
The present tool is implemented in Python for its backend part, and should work correctly with Python 3.7 or higher.
Get the github repo:
git clone https://github.com/kermitt2/kish
cd kishIt is strongly advised to setup first a virtual environment to avoid falling into one of these gloomy python dependency marshlands - you can adjust the version of Python to be used, but be sure to be 3.7 or higher:
virtualenv --system-site-packages -p python3.8 env
source env/bin/activateInstall the dependencies:
python3 -m pip install -r requirements.txtFinally install the project in editable state
python3 -m pip install -e .The web application requires Angular installed on your local system, for this you need Node.js and npm package manager. Then install Angular and the javascript dependencies:
cd kish/app
npm install
The KISH API service is implemented with FastAPI and can be started as follow on default port 8050:
python3 kish/service.py --config my_config.ymlThe application is then available for sign-in at http://localhost:8050/app/sign-in.html.
my_config.yml is the configuration file for the application, you can start with the default config.yml, but setting at least an admin account is preferable. By default, at first launch, an admin account is created with login [email protected] and password administrator. You should set a proper admin account in the config.yml file:
# default admin super-user when not already existing, admin_email must be an email and password must be kept secret
admin: "[email protected]"
admin_password: "admin"The admin has then the rights to manage users and extra fonctionalities in the web application.
The KISH web application is implemented with Angular and can be strated as follow on the default port 4200:
ng serve
By default, open the application at http://localhost:4200
Once the service is started as described in the previous sections, the API root point is by default http://localhost:8050, so for instance:
> wget http://localhost:8050/aliveThe web service API documentations are available at http(s)://*host*:*port*/docs, e.g. for instance http://localhost:8050/docs, based on Swagger, and http://localhost:8050/redoc for ReDoc documentation style. These documentations offer interactive support to support test queries.
A default configuration is set in config.yml, but it has to be completed for setting new host/port for the services and other deployment options. Sensitive/secret information are managed currently in the configuration file.
KISH is distributed under Apache 2.0 license.
The documentation of the project is distributed under CC-0 license.
If you contribute to the KISH project, you agree to share your contribution following these licenses.
Contact: Patrice Lopez ([email protected])