URL's shouldn't give CodeReview false positives when they're not pointing to CodeReview

Zomis / Duga

Stack Exchange Chat bot

18 stars 6 forks source link

URL's shouldn't give CodeReview false positives when they're not pointing to CodeReview #62

Open retailcoder opened 9 years ago

retailcoder commented 9 years ago

[ ] Determine if a comment contains a URL (regex?)
[ ] Determine if rest of comment contains a match. If not...
[ ] Determine if URL is pointing to codereview.stackexchange.com - any other domain/subdomain that doesn't match that, is a false positive (this way our dear @Duga should keep tracking links to codereview.stackexchange.com/help/on-topic)

Zomis commented 9 years ago

Erhm....! codereview.stackexchange.com/help/on-topic

retailcoder commented 9 years ago

@Zomis what's the domain/subdomain of that url again?

codereview.stackexchange.com

hmm maybe I need to fine-tune the way I word specs...

retailcoder commented 9 years ago

Oh wow, I think I just pinged someone named Duga!

Zomis commented 9 years ago

This problem also exist for the "Programmers" website, and it is a really difficult problem. For programmers.stackexchange I scan for keywords such as "better fit", "place to ask", "belong", etc. and I have still not been able to eliminate false positives. While the same thing could be done for Code Review, we don't have as many false positives.

Vogel612 commented 9 years ago

Another idea might be to ignore references to specific questions. There is no necessitiy to catch things like:

[closely related](http://codereview.stackexchange.com/q/45918/489)

Zomis commented 9 years ago

Yup. I'd like to come up with a nice way of filter things here. It is also important to remember that a theoretical comment such as:

This question is a better fit for Code Review,
 see [this related question over there as well](http://codereview.stackexchange.com/q/45918/489)

Such comments should still be posted.