The post discusses the evolution of property recommendation systems at Expedia Group, evaluating whether a high click-through rate correlates with a high conversion rate. It outlines the iterative development of models, noting how optimizing solely for clicks can negatively impact conversions. Advanced techniques, including contextual data and generative AI, aim to improve recommendation personalization and accuracy. The exploration includes methodologies like item-to-item interaction matrices, multi-task learning models, and de-biasing strategies for better results.
Table of contents
Does High Click-through Rate Lead to High Conversion Rate?AbstractIntroductionProblem statement and initial solutionFindings and algorithm iterationFuture directions48 Impressions