Process Improvement: Finding similar, duplicate or related issues

[troychaplin]

During the Docs Team planning meeting on March 11th there was a comment and brief mention of an idea to improve the system by introducing a Git Action that aims to find similar issues based on summaries or other issue data. I was asked by @zzap if I would like to take this task on.

I've worked out what I feel is a good start point and would need input from members of the Docs Team on the direction of the action and what data, and how, it would be gathered.

Here is a repo with an example Git Action that does the following when a new issue is opened, or an existing issue it labelled with Check for Related:

When a new issue is opened or labeled with 'check for related', the action does the following:

1. Matches similar words in the issue title (e.g., "image", "images", "imaging" are treated as matches)
2. Removes common words and short terms (less than 3 characters)
3. Searches for similar issues by matching related words
4. Requires at least 2 matching words (or 1 if the title only has 1 significant word)
5. Organizes matches into two groups and adds them as a comment to the issue:
   • Version-specific issues (with version labels like 6.7, 5.6)
   • Other related issues

An example of a newly opened issue with the comment can be seen here.

--

I'm not sure if this action in its current state is what the Docs Team will find beneficial. My hope with opening this ticket is to create a conversation around what the team needs would be, and what further modifications I could work on to meet those needs.

[github-actions[bot]]

Heads up @WordPress/docs-issues-coordinators, we have a new issue open. Time to use 'em labels.