摘要
CP-nets(条件偏好网)是定性表达偏好关系的一种图形工具,作为一种表达能力的工具,CP-nets功能强大,能直观、自然地表达用户的偏好信息。但是对于CP-nets学习的研究还不够深入,在实际应用中,由于用户行为或者观测误差的随机性,可能导致数据集中存在噪声数据,使得许多传统的学习方法无法得到最优的CP-nets结构。本文提出基于启发式算法的学习方法来解决CP-nets的结构学习问题。与传统方法中直接学习CP-nets结构不同,本文将CP-nets的结构学习问题转化为寻找最短路径问题,利用启发式算法的能力来寻找最优的CP-nets。
CP-nets (Conditional preference networks) is a graphical tool for qualitative expression of preference relations.As a expressive tool,CP-nets is powerful and can express user preferences intuitively and naturally.However,the study of CP-nets learning has not advanced sufficiently.In practical applications,due to the randomness of user's behavior or the observation error,it may lead to noisy data in data sets,which makes many traditional learning methods fail to get the optimal CP-nets.In this paper,a learning method based on heuristic algorithm is proposed to solve the structural learning problem of CP-nets.Compared with the direct learning CP-nets in traditional method,this paper transforms the structural learning problem of CP-nets into finding the shortest path problem,using heuristic algorithm to find the optimal CP-nets.
作者
仲兆琳
信统昌
ZHONG Zhaolin;XIN Tongchang(School of Computer and Control Engineering,Yantai University,Yantai Shandong 264000,China)
出处
《智能计算机与应用》
2019年第3期100-102,共3页
Intelligent Computer and Applications
