基于证据理论的不确定可拓推理及应用

An Uncertain Extension Reasoning Algorithm based on Evidence Theory

针对可拓推理的不足及可拓知识的不确定性,将关联函数与信任函数结合起来,研究不确定条件下可拓推理的可信度与支持度、冲突度与一致度,构建了基于证据理论的可拓推理函数及合成规则,以实现决策对象在不确定条件下的比较与选择,解决了多方案可拓推理、动态识别的数据挖掘问题,提高了可拓推理的准确性和可信度。

For the weakness of extension reasoning and the uncertainty of extension knowledge,we study the extension association and similarity, support and believability of extension reasoning under uncertainty by combining the correlation function and the belief function. We then build the extension reasoning functions and the combination rules based on D-S evidence theory. Not only does this method improve the accuracy and the reliability of extension reasoning by making comparison and selection of objects in changing environment, it also solves the problems of multi-project extension reasoning, and dynamic recognition in extension data mining.