Users are discussing how Discourse’s Suggested Topics algorithm can be improved to show topics more relevant to the current topic being read, rather than unrelated ones. LaurenN asks if relevance-based suggestions are possible or if limiting suggestions to the current category is the best workaround. sam invites feedback on the current algorithm on Meta. stance455 inquires about enabling both Related and Suggested topics on their forum. JammyDodger notes that displaying both sets is an upcoming feature. Later, JammyDodger confirms that Discourse AI now supports Semantic Related Topics, which uses semantic similarity to recommend more contextually relevant topics—especially useful for research—replacing random or unread-topic-based suggestions. This feature is already enabled on Meta for all users, including anonymous ones, and also powers semantic search results. The goal is to improve user experience by surfacing more meaningful next-read suggestions based on content relevance rather than randomness or category boundaries.
Hi all,
Considering how Discourse determines what Suggested Topics to show, is there a way to also determine suggested topics based on relevance to the current topic the user is reading? Instead of showing topics that are completely unrelated to the topic I have clicked on? Or is best practice to limit suggested to category?
Thoughts on current algorithm on meta?
On meta I notice Related and Suggested topics.
Is there a setting I need to enable so I can display both sets of links on my forum?
That’s actually an upcoming feature that should be available more widely soon. 
Just to round this out - Discourse AI now has the Related Topics available: 