The goal of INUUSE is to incentivize users to be engaged in interaction with items using a platform, eventually increasing their true reward and with incentives for expected utility.

To reach this objective the project will investigate the use of deep reinforcement learning methods exploiting historical users’ trajectories towards learning recommendation policies.

The recommender shall exploit:

• The interaction context: What are the users “trajectories”, what is their score, what items are those that can feasibly be recommended, data about the user.
• Past interaction experience: What are the items that other “engaged” users with similar context at a particular time have interacted with.
• Expectations and their probabilities: What is the probability of an expected reward, and how far is the prediction horizon for that expectation?
• Incentives: When an actual or expected incentive is to be provided and in what terms?