Section 6.7 and 6.6 The basic idea behind a monolith Recommendation System is to combine features from one recommendation system and use it as features for another Recommendation System.
Section 6.7 1. Meta-Level Hybrids: The model learned by one recommender is used as input to the next one. eg, build a user-word matrix for a restaurant recommender system and use this this to compute the the peer group and use that info along with a Collaborative methods to build one final Recommendation System.
Section 6.8: Feature Combination Hybrids: The basic idea is to combine the input data from various sources into a unified representation and use an existing technique to build a Recommendation System. For example: Augmenting the ratings matrix by adding columns for keywords obtained from knowledge-based systems.