Intelligent Software Agents

Probabilistic Data Mining
Methods for processing data have developed more slowly then methods for gathering and storing it. Hence, automated and semiautomated data processing tools are urgently needed. A promising new technique is based on learning Bayesian networks from data. Bayesian networks are graphical representations of the joint probability distributions for sets of variables. Currently, they play crucial roles in expert systems, diagnosis engines, and decision support systems. Many off-the-shelf tools can apply these learned networks; also, adaptive Bayesian networks offer semantic clarity and understandability by humans, ease of acquisition and incorporation of prior knowledge, ease of integration with optimal decision-making methods, the possibility of causal interpretation of learned models, and automatic handling of noisy and missing data.