(קבוצתיות-) הגבלות על מילים שנצפו

סיכום מבוסס בינה מלאכותית

In a discussion about [Group-]Restrictions on Watched Words, cuBEar highlights a potential privacy risk: while watched words can warn internal users about mentioning sensitive customer names, public users might also trigger warnings — inadvertently revealing confidential relationships. They suggest restricting watched words to specific user groups as a solution. sam responds that such granular business rules are too complex for core Discourse, but notes that the upcoming discourse-automation plugin (developed with @j.jaffe ux) may eventually support this. While group-scoped watched words aren’t planned for core, cuBEar appreciates the insight and expresses interest in following the automation plugin’s development. The thread reflects a common tension between out-of-the-box functionality and enterprise-specific needs, with plugin extensibility positioned as the pragmatic path forward.

Hi there!

I really like the watched words functionality.
However, I’ve stumbled upon a possible use case in which they on the one hand do what I expect, but a hidden risk is opened on the other hand.

Here’s my use case:

My company is working with customers whose names cannot be made public. So I’d like to implement something to “protect” my colleagues of mentioning those names in public discussions by accident.

So I could enter those names in the watched words, and people get warned in discussions - all fine, right?.

But: if some public user (so not a direct colleague) at some time by random occasion types in a customer’s name, they also get warned and thereby can maybe deduce a collaboration between us and that customer.

A solution to this could be restricting the watched words to a certain group of users.

Is that maybe possible somehow? :slight_smile:

I’d be very happy for any tips or hints!

2 לייקים

At some point the complexity of business rules gets a bit too much to include in core.

For cases like this we are working on a discourse-automation plugin with @j.jaffeux. This is certainly something automation could eventually support.

Groups scoping for various watched/flaged etc. words though I not something on the roadmap for core.

2 לייקים

Thank you for your answer, @sam !

I of course totally understand that core cannot cover every use case or business rules someone has :slight_smile:

Just wanted to see if there is something like a plugin or a nifty alteration I can do.

Will watch out for your plugin, thanks for the hint!