The Domain Knowledge Uncertainty Reasoning

As a new field of current research, the domain knowledge not only can significantly enhanced the memory capacity of work, but also improve the forecast ability for information in the individual cognitive. Ontology as an important knowledge model modeling tools, it is not only be described in the hierarchy of semantic and knowledge, but it also provides the standardization and clarification describe for relative concept. And it laid the foundation for the knowledge sharing, provides an excellent platform for the organization and establishment of domain knowledge. But the ontology can't well expression the degree intersection of the concept and the reasoning of only know the concept's part information. So the uncertainty problem exists in the domain knowledge has become the focus of attention and study. Bayesian decision theory prepared a strong theoretical basis to deal with various uncertain events or reasoning. In expression the uncertainty and reasoning side of domain knowledge, Bayesian Network was be widely used and has been proved the one of the most effective method to get some uncertainty knowledge's confidence. Conditional event algebra as a new subject has strong mathematical theory basis in dealing with uncertainty, probabilistic and ambiguity logic problems. It can tansformed the high-level conditional event into the conjeciton of ordinary event and logic event. This article combined the Bayesian Network and conditional event algebra to study the uncertainty event or knowledge exists in the domain knowledge.First of all, the article introduced some relevant technical of ontology. As a carrier of domain knowledge, the ontology can intuitive provide an mutual understanding for domain knowledge, clear the areas' identity words, and clearly defined this vocabulary (Terminology) and the relationship of this vocabulary. This article also introduced the build process and describe language of the ontology. To expand the corresponding probability of the ontology, make it have the ability to expressed uncertain information, eventually constructed domain ontology with probability information.Secondly, study the uncertainty reasoning based on Bayesian Network. Bayesian Network use conditional independence assumption between the random variables, straightforward expressed a joint probability distribution as a graph structure and a series of conditional probability table, after the corresponding variable elimination, it can calculated the probability distribution of the variable or a part of the variable. One of the most important reason of Bayesian Network are widely used for uncertainty reasoning is that the probability theory is a reasonable way to express the uncertainty. This article use the Stanford University's ontology development tool protege3.3.1 to building ontology, on this basis, to expanding the probability of ontology, to analyze the expanding probability ontology through the jena and related components. Then formatted the analyzed ontology file, and further to complete the construction of Bayesian Network. Last, by the constructed Bayesian Network, use the graphical way to achieve uncertainty reasoning process of the domain knowledge.Finally, the uncertainty reasoning which based on conditional event algebra and combined with the Bayesian Network is discussed. As a new subject, Conditional event algebra in uncertain information reasoning has a wide application space. Use conditional event algebra and conditional events, by extending the ordinary measurable space, first of all, make the probability and logic to be consistent in expression rules, then through the conditional event algebra, converted the high-level conditional event into the portfolio of ordinary event and logic event, and completed the reasoning process of high-level event. By example, this article proved the feasibility and effectiveness of this method to solve uncertain information of domain knowledge.