First, thanks to Github developers for topics, they are a great way of discovering projects and refining search!
There is one thing I lack about them which is some sort of topic normalisation/unification.
Some topics are multiword and I always strugle when choosing which topic to use in such case. E.g. imagine you implement a bot for reddit: you could either reddit-bot or redditbot tags.
And indeed, if you search for them, both give different results:
My suggestions would be
- change topic search algorighm so hyphens/underscores (or perhaps any special characters) are ignored. Worst case it would add few false positives (although I’d struggle to come up with example)
- suggest topic statistics when you add topics to your repositories. Then perhaps ambigous tags would happen less because the most common topic (reddit-bot in this case) would be suggested first. Here’s how Stackexchange does it for inspiration: