摘要
指出大数据时代的到来使自动分类再次受到人们的关注。总结现有的自动分类方法,介绍中国科学院文献情报中心的KOS引擎项目中的集成知识组织体系。在此基础上,改进BP神经网络算法,提出NIKOS自动分类模型。最后,通过实验检验基于N-IKOS分类的准确性,通过基于BP神经网络的分类实验、基于KOS引擎的分类实验和基于N-IKOS的分类实验比较新模型在自动分类中的优劣。实验结果表明:该研究改进了原有的KOS引擎分类,可为自动分类领域提供新的思路。
Automatic classification is taken attention again with the coming of big data. The paper summaries the methods of automatic classification, and introduces the integrated knowledge organization system in KOS engine project of National Science Library. Then, it improves the BP neural network, and raises a pattern of N-IKOS automatic classification. In the end,the paper tests the accuracy of N-IKOS automatic classification by experiment, and compares the merits and drawbacks of the new model with the experiments of automatic classification based on BP neural network KOS engine and N-IKOS. It improves the category of the KOS engine classification, so as to provide the new thought for automatic classifica- tion research.
出处
《图书情报工作》
CSSCI
北大核心
2014年第24期106-112,共7页
Library and Information Service
关键词
自动分类
知识组织体系
机器学习
BP神经网络
automatic classification
knowledge organization system
machine learning
BP neural network
