In this talk we describe a computational framework for modelling data generated by complex processes. Specifically, we review the framework of probabilistic networks, emphasizing basic computational issues and applications. The research problems that we discuss include the specification, automatic learning, and manipulation of probabilistic knowledge. We sketch several novel algorithms for querying and updating probabilistic networks and their applications to biological data modelling. We also discuss the application of probabilistic networks for designing effective memory-based reasoning systems. We conclude by surveying the interdisciplinary applications of these intelligent modelling tools.