Often technologies such as fault-trees, flow-charts, simple wiki-based web sites, and other document-based approaches are used for collecting and sharing expert knowledge and experience. The problem, however, is that we often lose our way in the myriad of information when trying to search for answers that can help us when trying to fix a problem as there are no dynamic elements that can help us navigate or utilize equipment state to increase precision. Another problem is that it really isn’t knowledge because there is no common interpretation of all this information and people just understand complex information in different ways. I might jump to one conclusion and you might another – so how can we apply the same interpretation an expert would within a certain problem area in a consistent way regardless of the skills of the user? How do we capture this important “know-how” and distribute it to other people for use in troubleshooting?
Bayesian Networks are a fundamental reasoning technology particularly suited for handling complex and uncertain domains. A Bayesian Network is a model of the domain where events and their relationships are represented in a graph with associated probabilities.
The Dezide decision engine executes Bayesian Network models in a highly efficient manner, finding an optimal sequence of steps balancing belief of the step being helpful with its cost. The Dezide engine produces a sequence of solution-oriented or information-gathering steps and considers many factors.
The below table compares various types of technologies used by vendors of AI technologies for technical troubleshooting concerning complexity and performance when used in different situations.