Problem with any recommendation feature is ...
How is it actually going to sort the novels? What data is it based on?
It couldn't be based on tags, because ...
That's what search function already does. You could search by tags, and it is ordered based on the ranking of the said tag.
It could be random, which isn't hard to code, but it would provide the stories you wouldn't like.
Actually, if it provides the story you didn't like, the reader's first instinct is to rate is 1* or whatever is lowest available to "teach the system a lesson". It buries the story for everyone else, and system doesn't learn anything from it. The Royal Road, with practically infinite money, still didn't figure out how to teach readers about the existence "do not show me this" button.
I couldn't think of any way how to code such recommendation function.
Well, you could feed it to the "AI" ChatGPT equivalent, and then simply ask, but this is counterproductive. So many people rage against the "AI" use, and you are going to consciously feed the stories to the "AI"? This is the exact opposite of what was attempted so far. Not to mention, this is not going to work - the "AI" is bad at comprehending text.