Abstract
Recommender Systems (RSs) generate personalised suggestions for items and can influence collective users' choices behaviour. The impact of operational RSs on users' decisions can be assessed by analysing the actual choices' diversity, and quality, e.g., the users' satisfaction for their choices. But, in order to estimate the potential impact of an RS in new scenarios and for not yet deployed RSs, simulating user-system interactions can be valuable. We here illustrate a simulation framework consisting of users, items, and alternative RSs. We simulate users' choices over consecutive time intervals, by assuming that an RS influences the users' choices with recommendations. We measure global properties of the simulated choices, such as their diversity and quality. The obtained results, and the proposed simulation framework, can be used by a system designer in order to anticipate the effect of a candidate RS in its long term usage.