FM-Intent: Predicting User Session Intent with Hierarchical Multi-Task Learning

The blog post introduces FM-Intent, a novel recommendation model developed by Netflix that employs hierarchical multi-task learning to enhance user intent prediction and next-item recommendations. By capturing both short-term and long-term user interests, FM-Intent significantly improves recommendation accuracy, outperforming existing models. The post details the model's architecture, including user intent prediction using a Transformer encoder, and presents experimental results that validate its effectiveness.