Section 16 of 24 · 5 lessons
Recommendations and Ranking
Botie has more guides than a reader can inspect. We will work out which ones to suggest using popularity, descriptions, and shared preferences, then evaluate the short list the reader actually sees. Missing interactions and new readers need their own treatment.
Choosing Useful Candidates
A reader should not have to search the whole shelf. Begin with what previous readers opened, then use guide descriptions when interaction history is missing.
Learning From Interaction Patterns
Readers can share preferences that item descriptions miss. Keep missing ratings distinct from dislikes while learning from neighbours and compact factor representations.
Evaluating Recommendations
A useful guide needs to reach the part of the list someone sees. Evaluate the visible ranking and inspect readers and items with little history separately.