Every magician needs a beautiful assistant!
Cannif is a web GUI for the Annif toolkit. It provides:
- a sortable, searchable, export-able list of projects and details
- an editor for projects and backend parameters
- simple charts of evaluation metrics
It is a pretty small script built using Streamlit.
For Annif installations of a few projects, editing projects by hand and managing them isn't too much work. However, for cases where there are tens of projects or more, it quickly becomes onerous. Additionally, gathering the output of dozens of evaluations and keeping them organized can be a pain. These things aren't the job of Annif, so cannif was created.
There is some discussion about the relationship between the Annif web UI and cannif at this issue.
Why cannif? In Canada we often prepend 'can' to names, so can-Annif, cannif. In French, canif means pocket knife (with an interesting history), which is a very useful tool indeed.
The recommended way is to install cannif into a virtual environment.
python3 -m venv cannif-venv
source cannif-venv/bin/activate
pip install annif
pip install streamlit
Open a new terminal. Start up the application:
streamlit run cannif_streamlit.py
You should see:
You can now view your Streamlit app in your browser.
Local URL: http://localhost:8501
Network URL: http://192.168.1.107:8501
cannif will automatically try to start an Annif REST server running at http://localhost:5000/v1/. cannif looks for project information in the default location, projects.d. If cannif is run from the Annif folder, it will use the projects defined there.
When connected, it presents an interactive table of projects and their details. The table can be downloaded as a CSV file.
Projects can be created through the "New Project" pop-up. A project requires:
- a name
- a loaded vocabulary
- a language
- an analyzer
- a backend
Vocabularies can be loaded for a project by entering a vocab ID and a language, and uploading a vocabulary file. Once loaded, the vocabulary can then be used in any new project. Annif will reload when a new project is created.
When you select a project in the table, details about that project and its backend appear below the table.
Updating project configuration is coming!
Projects can be trained by uploading a training corpus file. Training is started in the background; training may take a long time and use a lot of resources. When a project is trained a ✔ will appear in the table.
Projects can be evaluated by uploading an evaluation corpus file. Evaluation is started in the background. When a project has been evaluated, metrics will appear in both the project details and in a series of comparative charts.
Cannif looks for evaluation information in the data/eval directory. JSON files in the Annif --metrics-file format will be used if they match the name of a project_id.
Tasks running in the background are listed below the project information. When tasks are completed, their status (finished or failed) is shown along with the relevant output. Tasks can be cancelled by clicking on the 'X' next to their process ID.
If you have any other ideas for how to improve cannif, get in touch or submit an issue!