论文部分内容阅读
探索数字图书馆的图书、期刊、网页等不同类型的馆藏文献进行分类组织时,多种类型文献单独及混合自动分类的特点。采用KNN等分类算法对多种类型文献进行分类实验并分析比较。仿真实验表明:随着实验文本数增加,各种类型文献的单独自动分类准确率有2%至7.8%不等的提升。图书与学术性期刊、网页与非学术性期刊之间的混合分类效果较好,分类准确率可达到85%以上;对多种类型文献实现基于自动分类的统一组织具有较高的可行性。
Explore the digital library books, periodicals, web pages and other types of collections of documents classified organization, a variety of types of documents alone and mixed automatic classification characteristics. Using KNN classification algorithm to classify different types of literature experiments and analysis. The simulation results show that with the increase of the number of experimental texts, the accuracy of individual automatic classification of various types of documents increases from 2% to 7.8%. The hybrid classification between books and academic journals, web pages and non-academic journals is better, and the classification accuracy rate can reach above 85%. It is feasible to realize the unified organization based on automatic classification for many types of documents.