| The Bayesian network is a concise graphical presentation of a knowledge system under uncertainty. Evaluating influence diagrams using Bayesian network inference is becoming attractive for the Bayesian network has strength in inferencing under uncertainty and many efficient Bayesian network algorithms have been proposed in the past decades.; In the first part of this thesis research, we propose and implement one simple method to evaluate influence diagrams using Bayesian network inference, which we claim and prove is more efficient and simpler than previous methods. In the second part of this thesis research, we implement an application of shop assistance whose decision module is represented as influence diagrams. We describe the architecture of the shopping assistant and some important component implementations. We analyze some potential problems we must solve when we develop our system. |
