How to properly implement an AI-based recommendation system?

Question

Machine learning

There is a task to predict and recommend courses to people based on the data they provide (user profile - what do you want, what goal are you pursuing, what do you do?) and the statistics collected on the number of courses taken by other people. The machine must constantly study the received data and improve the quality of the prediction.
After doing a little research, I found several frameworks for js and python. As I understand it, the difference in stack choice is only a matter of performance. The most important task is to understand on the basis of which formula and representation, or graph, to implement the function of calculating the input data, in order, firstly, to translate text markers, such as course categories, tags, and user-entered data into numerical values, and secondly, to compare them how close they are to the desired category of courses. Are there any references, manuals that will lead to understanding, or maybe someone who is an expert can sort it out, point out landmarks?

Answer 1

Throw away AI. You don't really need it.
Implement a questionnaire, for each answer a score is added to some specialty.
As a result, 5 top specialties are displayed.
Why complicate things?

Answer 2

Manuals for creating recommender systems - a little less than dofiga, you just need not be too lazy to type 'python recommender system' in Google.
To at least get into the topic a little, free courses are enough, for example: https://stepik.org/course/4852/syllabus

Answer 3

Grokaem algorithms, there is a simple example.
Or "Recommender systems in practice | Falk Kim"