marks or suggest related Issues about ideas HOT 6 OPEN

I also want assistance managing numerous duplicate issues for my project https://github.com/Glavin001/atom-beautify

I developed https://github.com/Glavin001/issuebot with this goal in mind. I'd like to rebuild it using Probot.

from ideas.

Check out https://github.com/probot/duplicate-issues as a starting point.

There's an instance of it running at https://github.com/apps/duplicate-issues. The app still needs some tweaking and improvement, which is why it's not listed on the Probot site yet.

GitHub

probot/duplicate-issues

duplicate-issues - A probot plugin for automatic issue duplicate detection

GitHub

Build software better, together

GitHub is where people build software. More than 25 million people use GitHub to discover, fork, and contribute to over 70 million projects.

from ideas.

There's an instance of it running at https://github.com/apps/duplicate-issues. The app still needs some tweaking and improvement, which is why it's not listed on the Probot site yet.

Where exactly does the app need to be improved? Are there any examples where the app doesn't work as it is supposed to? @bkeepers

GitHub

Build software better, together

GitHub is where people build software. More than 28 million people use GitHub to discover, fork, and contribute to over 80 million projects.

from ideas.

Where exactly does the app need to be improved?

• The matching algorithm is very crude. Ideally it would to use some more sophisticated machine learning, but I don't exactly know what that would be
• Performance is not great because it doesn't persist the training data, so it has to fetch all issues whenever a new issue is open

from ideas.

The matching algorithm is very crude. Ideally it would to use some more sophisticated machine learning, but I don't exactly know what that would be

So no fuzzy searching? There are some options like https://www.npmjs.com/package/string-similarity and https://www.npmjs.com/package/similarity but they don't use complex ML algos. I don't think we need ML for this application.

Performance is not great because it doesn't persist the training data, so it has to fetch all issues whenever a new issue is open

We can maintain a issues array and whenever a new issue is opened push that issue into the array so that all the issues need not be fetched everytime?

from ideas.

Does anyone know if that bot or something similar is now functional?

from ideas.