https://github.com/pchaigno/trieur

Automatically classify your mails into the right folders. Test version.
https://github.com/pchaigno/trieur

Last synced: about 1 month ago
JSON representation

Automatically classify your mails into the right folders. Test version.

• Host: GitHub
• URL: https://github.com/pchaigno/trieur
• Owner: pchaigno
• License: mit
• Created: 2015-05-31T09:18:51.000Z (over 9 years ago)
• Default Branch: master
• Last Pushed: 2015-11-15T17:14:42.000Z (about 9 years ago)
• Last Synced: 2024-10-13T22:47:23.600Z (2 months ago)
• Language: Python
• Homepage:
• Size: 340 KB
• Stars: 1
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Trieur

Ultimately, this project will automatically classify your mails into the right folders.

Right now, I need help to validate my first simple implementation.

## Usage & Validation

0. Retrieve the project: `git clone https://github.com/pchaigno/trieur && cd trieur/`.

0. Install the dependencies: `sudo apt-get install gfortran libopenblas-dev liblapack-dev && sudo pip install -r requirements.txt`

0. Download your emails: `./imap.py imap_server path/to/temp/folder`.

0. Test the Bayesian classifier: `./train.py path/to/temp/folder`.

0. Test the SVM classifier: `./train.py -c SGDClassifier path/to/temp/folder`.

0. If the accuracy computed through the tenfold cross-validation is below 90% please open an issue with your results and the name of your mailbox folders.

## How does it work?

`imap.py` downloads emails from your mailbox folders. `train.py` then creates a classifier with the given algorithm (e.g. Bayesian) and tunes it (computes the best parameters by cross-validation). Finally, it displays the result of the tenfolds cross-validation.

## Frequently Asked Questions

Please see the [Frequently Asked Questions page](https://github.com/pchaigno/trieur/wiki/Frequently-Asked-Questions) for answers to your questions. Don't hesitate to open an issue if your question is not answered in the Wiki.

## License

This project is under [MIT license](LICENSE).