Bayesian Inference : Probabilistic evidence for User Modeling
Rebai Rim, Ali Ben Mrad, Maalej Mohamed Amin · Procedia Computer Science · 2025
The application of Bayesian statistical approaches has grown in the last few years in a variety of scientific domains, including computing. Bayesian inference integrates experimental data with prior knowledge to perform analysis. Unlike classical methods, Bayesian approaches offer fundamentally different perspectives on statistical inference. In this paper, we adopt Bayesian inference as a thorough and reliable formalism to handle uncertain data in user modeling. Inference methods are driven by evidence. Hard evidence and probabilistic evidence are the two main categories of evidence that are the subject of this essay. Specifically, we emphasize updating the evidence represented in a Bayesian model. The paper illustrates the use of fixed probabilistic evidence, or soft evidence, in an adaptive user interface.