论文部分内容阅读
“知识编译”这一术语是指:给出某知识K 和一个知识编译函数 C,将 C 作用于 K 得出编译知识 K’;即:C(K)=K’。假设对完成某一信息处理任务 f,K’比 K 有效,即:f 是期望实现的某种功能,而且利用 K’实现 f 的某功能(以 f_k,表示)要优于利用 K 完成 f 的任何功能(以 f_k 表示)。由 f_k 与 f_k 实现的功能也许不同于 f 的功能。例如:f 是诊断某一设备的任务,K 是关于此设备的一个定性物理模型,C 的功能是将 K 转化为诊断知识 K’。换言之:K’比 K 更直接地将观察与故障相联系。只利用 K 得到一个诊断系统的实现过程为 f_k,而利用 K’的诊断系统实现过程为 f_k’。
The term “knowledge compilation” means that a knowledge K and a knowledge compilation function C are given, and C is applied to K to obtain compilation knowledge K ’; ie, C (K) = K’. Suppose that for some information processing task f, K ’is more efficient than K, that is: f is some function expected to be realized, and some function of f (using f_k) is better than k Any function (denoted by f_k). The functions implemented by f_k and f_k may differ from the functions of f. For example: f is the task of diagnosing a device, K is a qualitative physical model of the device, and C’s function is to convert K into diagnostic knowledge K ’. In other words: K ’associates observations with faults more directly than K. Only using K to get a diagnostic system for the realization of the process f_k, and the use of K ’diagnostic system for the realization of the process f_k’.