摘要
kNN算法用已归类的数据训练分类器,它是一种基于实例研究(instancebased learning)文本分类算法.本文在研究kNN算法的基础上,结合k邻近法和最近特征线法的思想,提出了新的分类方法,k最近特征线法(k nearest feature line,kNFL),将其运用于文本分类中,汲取了kNN算法和NFL算法的优点,降低了偶然误差,提高了算法适应性和分类精度.
kNN （k nearest neighbor） algorithm use labeled data to build the classifier. It is a instance-base learning algorithm. In this paper we will try to improve it on the basis of researching the kNN algorithm. Propose a novel classifier, kNFL, which combines the advantage of the kNN algorithm and the NFL algorithm, it reduces accidental error, and advances the algorithm＇s flexibility and classifier＇s precision when it is applied to the text categorization.
出处
《南华大学学报（自然科学版）》
2005年第3期78-80,共3页
Journal of University of South China：Science and Technology
