Agentic-Priors: Using LLMs to Address Cold Start in Preference Elicitation
What it is
This paper introduces Agentic-Priors, a method that uses Large Language Models to generate interpretable priors for user preferences (e.g., willingness to pay) to solve the cold start problem in choice-based systems. It maps these LLM-derived trade-offs to a prior over multinomial logit coefficients, which is then refined by a Bayesian learner as real user choices are observed.
The problem it solves
The cold start problem in preference elicitation, where agents lack historical choice data for new users and must rely on uninformed or poorly fitted priors.
What you could use it for
- Personalizing recommendation systems for new users with no interaction history
- Optimizing pricing or service options for new customers in ride-sharing or travel platforms
- Initializing preference models in healthcare or financial decision support tools
Read the source ↗Back to the library →
Summary written by an AI model from the source; check the source before you rely on it.