Neuroscientific User Models: The Source of Uncertain User Feedback and Potentials for Improving Recommendation and Personalisation

TL;DR

This paper leverages neuroscientific theories to model user behavior variability as informative data, aiming to enhance recommendation systems by translating neural activity into decision-making insights.

Contribution

It introduces a novel neural-based user model that transforms behavioral variability into valuable information for improving personalization and recommendation accuracy.

Findings

Neural models can effectively estimate user feedback.

Decoders translate neuronal activity into decision metrics.

User clustering based on neural traits improves collaborative filtering.

Abstract

Recent research revealed a considerable lack of reliability for user feedback when interacting with adaptive systems, often denoted as user noise or human uncertainty. Moreover, this lack of reliability holds striking impacts for the assessment of adaptive systems and personalisation approaches. Whenever research on this topic is done, there is a very strong system-centric view in which user variation is something undesirable and should be modelled with the eye to eliminate. However, the possibilities of extracting additional information were only insufficiently considered so far. In this contribution we consider the neuroscientific theory of the Bayesian brain in order to develop novel user models with the power of turning the variability of user behaviour into additional information for improving recommendation and personalisation. To this end, we first introduce an adaptive model in which populations of neurons provide an estimation for a feedback to be submitted. Subsequently, we present various decoder functions with which neuronal activity can be translated into quantitative decisions. The interplay of cognition model and decoder functions lead to different model-based properties of decision-making. This will help to associate users to different clusters on the basis of their individual neural characteristics and thinking patterns. By means of user experiments and simulations, we show that this information can be used to improve the standard collaborative filtering.