dmonego / hacktoberfest-issue-classifier

Hacktoberfest is great, but the issues are full of spam. This is a tool to classify issues to help find good ones to work on.

Issues are put in three categories:

• Good: Meaningful work on useful projects.
• Spam: Churn for Hacktoberfest T-Shirts. Empty repos, write fizzbuzz in 30 languages, meaningless junk.
• Novelty: Real, working projects with just for fun tasks. Participatory art, GPT3 poetry, Hacktoberfest stuff with real goals and engaged maintainers.

hacktoberfest-issue-classifier's Issues

Classify dataset set-1

Classify a dataset of hacktoberfest issues as spam, good, or novelty.

To start the classifier, clone the repository and cd into the directory.

Then, set up the repository and start classifying:

git checkout -b classify-set-1
make venv
. venv/bin/activate
jupyter notebook classify.ipynb

Once you're in the notebook, set the dataset to set-1. Hit run, and then hit run again. You'll be presented with an issue - take a look, and classify it as good, spam, novelty, or skip it. Repeat until the set is done, then run the last block to save the results.

Then, add the information at datasets/set-1/annotated.json to git and check it in

For guidelines:

• Good issues are interesting issues on useful projects
• Novelty issues are things like participatory art or self referential hacktoberfest stuff with a real project and engaged maintainers.
• Spam issues are empty issues or projects with no real use outside of churn for hacktoberfest T-Shirts

This classification data will be used to train bots to filter issues.