Part 3 Reflected intelligence
Search ranking shouldn’t be a static function. Every query and user interaction with search results is a signal that can be used to improve the relevance of future results. In the last part, we learned domain-specific knowledge from content and signals, and we used that knowledge to interpret query intent.
In this part, we’ll take a deeper dive into the topic of reflected intelligence, the process of learning from user interactions with search results to improve relevance ranking models. We’ll extend our coverage of crowdsourced relevance algorithms from chapter 4, with chapters dedicated to three key types of reflected intelligence: popularized relevance (signals boosting), personalized relevance (collaborative filtering), and generalized relevance (learning to rank).