There might be a useful overlap with transliterature technology here.

John Ohno on 2018-12-24

There might be a useful overlap with transliterature technology here. Translit provides several useful things: permissionless remixing with automatic attribution (through transclusion), monetization of the original contexts in proportion to the material borrowed (through transcopyright), and open contextualization (through transpointers / links). Implementations should typically put together / pay for segments only when they become visible, in order to avoid scams.

The only major aspect of your model not addressed is the ability for professional recommenders to get paid based on pages read — something that might be a little tricky, but could be solved (in a hacky manner) by each recommender creating a new edition of a recommended work with content they own injected on each page (sort of like how an affiliate link works).