Problem with meaning in texts generated by AI?

Question

Algorithms

I decided to create an application that will write fairy tales based on old Russian fairy tales. I decided to use Markov chains for this task, but this resulted in a new problem: the text becomes meaningless after processing. Having searched the Internet for articles about writing texts by artificial intelligence, I did not find any description of the logic of creating a logically complete story. Please help in any way you can.

Answer 1

See here for a list .
You need to be able to do the same: Demo

Answer 2

Markov chains always generate random data (garbage), just limited by certain rules (probability sizes).
You can raise the quality using competitive GAN approaches (in the field of generation, this approach is now at the forefront).
more details here:
https://en.wikipedia.org/wiki/%D0%93%D0%B5%D0%BD%D...

Answer 3

Well, we still don't have AI. There are some neural networks that work on certain tasks. But the main problem for you is that neural networks do not understand the meaning of the text.
So, the task is not feasible at the moment. For this, the necessary “neural network”, which grew up listening to fairy tales from my mother in childhood, went through the stages of personality development, filled “bumps”, so to speak.
It's easier to hire a rewriter - a person.