实际问题中存在着大量连续量以及对它们的各种认知信息(知识).为了较好地描述和利用这些知识,将这些知识分解为连续认知结论以及关于这个结论的一个认知结构,并将它们统一成一个连续认知结构.在此基础上提出了一种基于连续认知结构的连续认知推理网络,并探讨了一套利用连续认知结构进行推理的方法.这种方法在不完全的知识推理网络中的推理是非单调的,其最简推理复杂度关于推理网络节点在最复杂情况下是线性的.同时,还用例子说明了这种方法在实际问题求解中的适用性.

There are a lot of continuous values and various kinds of cognitive information (knowledge). Firstly, for describing and handling the knowledge, in this paper, they are divided into two parts: a continuous cognitive result and its cognitive structure. Secondly, a cognitive structure and a cognitive inference network are put forward, and they are integrated into a continuous cognitive structure. Thirdly, a continuous cognitive inferential network based on continuous cognitive structure is proposed. Fourthly,a set of inference approaches are discussed utilizing the continuous cognitive structure.The inference is non-monotonous upon incomplete inference network.The most simplified computational complexity of inference is linear is linear to the inferential nodes under the most complex condition.Finally,the suitability of the approach in practical problem is shown by an example.

国家自然科学基金资助项目(69875017)