Algoritme om content af te stemmen op gebruikersinteresses

AI-generated summary

The discussion explores whether Discourse can tailor content to user interests based on browsing behavior, similar to platforms like Reddit or Quora. Jnolo asks if Discourse can learn user preferences through interactions such as clicks, reading time, and engagement, then recommend relevant content. EricGT initially suggests Prolog, an AI language, but later clarifies that such a recommendation system is better approached with neural networks and AI, as in Netflix’s recommendation engine, rather than logic-based systems. The conversation highlights the distinction between open-world AI problems (suitable for machine learning) and closed-world logic systems (like Prolog), concluding that modern content personalization typically relies on machine learning rather than rule-based approaches.

It’s very apparent that Discourse is great for forums. Are there algorithms / capabilities to tailor the content shown to each user, based on their preferences (i.e. browsing history)?

Like Reddit tailors content based on what questions to click one, to show what’s most interesting to you…

1 like

Not everyone here may use Reddit, I don’t. However I do program in Prolog which is an AI language and this problem is probably something that can more easily be done using Prolog.

More details of how you envision this happening would be nice.

1 like

Or, like Quora (if you use that).

Basic idea, as a user navigates the site, the site learns what they are most interested in (by which posts they click on, and how long they read it, and whether they comment/like). Then, the site shows them more of what they like. Is this possible in Discourse?

Any other recommendations for what platform does this out-of-the-box/standard?

1 like

Nope.


OK.

I take it you want something like

How Netflix Uses AI For Better Content Recommendation

So in the world of AI I would classify your problem as an open world problem and use neural networks (thinks self driving cars) and not classify your problem as closed world and use Prolog.

1 like