Deep Learning for Search Relevance
Facebook Search is a product which helps people finding what the world is saying about topics which matter to them. Given that the index has more than 3 trillion posts and more than 2B highly personalized queries, the relevance (query <-> document) is a challenging problem. Through the application of deep neural networks for learning-to-rank methods, the need for extensive feature engineering was obviated.
Christina Scheau is an Engineering Manager at Facebook, with expertise in artificial intelligence, search engines and natural language processing.