摘要
针对数据挖掘中的文本自动分类问题,提出了一种基于加权模糊推理网络的分类方法.网络的基本信息处理单元为模糊推理神经元,融合了模糊逻辑能够较完整的表达领域规则和先验知识以及神经网络自适应环境的优点.根据模糊推理规则的量化表示形式和微分方程数值解的动力学思想推导出网络一种新的学习算法.该算法以文本特征谓词的真度作为分类依据,体现了模糊分类的思想.以旅游站点网页分类为例验证了该方法的有效性.
Aiming to document classification in data mining, a weighted fuzzy reasoning network structure model and its algorithm were proposed. The basic unit of information processing in this network is fuzzy reasoning neuron. The algorithm combined the excellence of fuzzy logic so that it could express domain rules and metempirical knowledge with the virtue of neural network, and it could automatically adapt to the environment. Learning algorithm of the network was shown based on fuzzy reasoning rule and numerical method for differential dynamic systems. This algorithm was executed according to the document feature predications, and it presented fuzzy classification characteristics. Finally an experiment example was presented to prove the availability of the model and algorithm by the classification of tour websites.
出处
《哈尔滨工程大学学报》
EI
CAS
CSCD
2004年第4期504-508,共5页
Journal of Harbin Engineering University
基金
国家自然科学基金资助项目(10172028).
关键词
数据挖掘
模糊分类
神经网络
学习算法
Data mining
Learning algorithms
Neural networks
Numerical methods
Websites
Weighing
