Aplicação de redes bayesianas a dois problemas atuais

Abstract

Problems found in various areas of knowledge may have solutions with the aid of mathematics and computing. In this work were addressed two problems. The first, the biological area, comes to a paternity test. The objective was to develop a model, and implement it in software to take account also situations where the supposed father is unknown. The second issue deals with the construction of recommendation systems. Papers found in the literature has demonstrated inefficiency of existing architectures for building these systems. It was proposed a model that minimizes the difficulties encountered. To approach the two problems, was offered the use of Bayesian Networks. These are graphs that express variables and probability distributions, and has capacity to deal with the uncertainty.