Currently under review as a demo at EKAW 2024. See also https://github.com/LiUSemWeb/entity-recontextualization.
To run the system, for now you will need to have Flutter and Python installed.
After cloning, navigate to the root of the project.
Build the web project using flutter build web --web-renderer html.
This should add content to lintext/build/web including, among other things, index.html.
With this running, you can start a webserver using python lintext/server/python/main.py.
The UI can be access by default at localhost:13679.
You can also run the webserver without building the web version of the UI.
Then you need to open/run the UI through other means, such as VSCode by opening main.dart and building/running the app specifically for your OS.
You likely want to run the python side of things in a virtual environment.
If you're using conda, you can use the .yml file provided to create the environment we use: conda env create -f lintext/server/relex.yml.
- Python 3.8
- uvicorn 0.29
- fastapi 0.115
- pytorch 2.4.1
- pytorch-cuda 12.4
- transformers 4.40.2
See lintext/pubspec.yaml for Flutter dependencies. We use flutter sdk version 3.5.0.
- The AV will let you select different similarity metrics.
- The DV will let you choose different tokenizers.
- The SV will include a select all/none button
- ... (^ High priority above, low below v)
- The AV will include a little color key that lets you change the low/high colors.
- Renaming new relation should instantly update the
*(AV = Analyze View, DV = Document View, SV = Schema View)