The recommendation tool seems to be able to provide suggestions similar to a given article. This can be used to support a predefined set of top-level categories based on the list of vital articles.
We can present a category for "Artists" and provide suggestions similar to one article randomly picked from that group (Leonardo da Vinci, Michelangelo, El Greco, Rembrandt, etc.).

That will avoid the complexities of filtering article recommendations based on their Wikipedia categories, but still providing users with control on which kind of suggestions they want.

In this comment a user highlights criteria considered relevant when selecting articles:

1. Not only featured articles, but also good.
2. Search articles from categories. For example sport, etc.
3. Depending on popularity. Number of views in a day. Not only the most popular, but as a parameter. For example more than 10, but less than 100 views in a day.

@Pginer-WMF, is this ticket and other that describe topics (and other related things) for list of suggestions still relevant?

In T113257#4020026, @Petar.petkovic wrote:

@Pginer-WMF, is this ticket and other that describe topics (and other related things) for list of suggestions still relevant?

Yes. From user research we identified users could take advantage of having more control on the suggestions they are provided. In particular, selecting a specific area of interest to discover topics they can translate seems relevant. The dashboard has evolved since then, and the designs need updates but the ideas are still relevant.

In this comment an editor proposes to provide the ability to focus on good and featured articles.

In this comment an editor asks for the possibility of customizing the suggestions ot pick more specific topics of interest:

Instead of guessing, can I just select the topic myself? For example, If i want some volcanoes, I could choose the topic "Volcanoes". If rivers, "Rivers" topic would be good. Maybe Wikipedia categories can help to choose a preferred topic for the article.

This issue is mentioned again by a user in this comment:

I mainly focused on computer science, I'd like to translate those pages, however on translation tool, it recommends me a lot of figures page, which I really can't help much

I just wanted to mention a use case for this work that relates to WMF's Product & Technology 2023-2024 Annual Plan. The Growth team is considering ways in which we can impact the WE1.3 Key Result:

1 percentage point increase (YoY) in the portion of newly created or improved articles on high-impact topics with acceptable quality, per the “global quality score”, that are created or edited on Wikipedia, starting with underrepresented geographic regions and gender.

One idea that has arisen is if we added a call to action from the Newcomer Homepage to help contributors identify and expand missing content across languages and topics.

Hypothesis: If we present Homepage editors with Content Translation tasks that are filtered to high-impact topics, then more articles about these topics will be created.

I think this would be an especially compelling idea if the Content Translation topics related to Suggested Edits topics, so Content Translation could be integrated into Suggested Edits (on wikis that are interested in having Content Translation tasks available to new editors).