Controllable Text Recommendation

Controllable text recommendation aims to build systems that provide personalized recommendations while allowing users to directly influence the results. Current research focuses on integrating large language models (LLMs) and leveraging techniques like reinforcement learning and counterfactual explanations to enhance user control and transparency, often employing bi-encoder retrieval models for efficient search. This field is significant because it addresses limitations of traditional recommender systems by improving user satisfaction, trust, and ultimately, the accuracy and relevance of recommendations across diverse applications, such as scientific dataset discovery and personalized content delivery.