维基百科在多种类型数字文本资源自动分类中的应用

发布时间：2018-02-21 21:08

本文关键词： 书目信息新闻网页自动分类维基百科特征扩展　出处：《情报科学》2017年02期 　论文类型：期刊论文

【摘要】：【目的/意义】书目信息和网页等不同类型文献之间存在特征词不匹配等语义差异问题,使得将书目信息作为训练集来对网络信息资源进行自动分类时,现有分类方法的分类性能不佳。【方法/过程】文章提出使用维基百科开展语义特征扩展,解决语义差异问题的自动文本分类方法。使用数字图书馆中容易获取类别标识及摘要等文本内容的书目信息作为训练集,引入第三方资源的维基百科对其进行语义特征扩展,缩小作为训练集的书目信息与作为待分类文本的网页之间的语义差异,对属于不同文献类型的网页进行分类。【结果/结论】实验表明与未经过扩展的分类方法相比,分类准确率分别提高5.5%至8.4%,证明该方法能够有效提高文本自动分类的分类效果。
[Abstract]:[feature words do not match the semantic differences between bibliographic information and web pages to different types of literature / significance], the bibliographic information as the training set for automatic classification of network information resources, the classification performance of the existing classification methods are poor. [method] / process is proposed in this paper using Wikipedia to carry out semantic extension, automatic text classification method to solve the problem of semantic difference. Category identification and abstract text content accessible using the digital library bibliographic information as the training set, the introduction of third party resources Wikipedia semantic features of its expansion, narrowing as bibliographic information and semantic differences between the text to be classified as "the training set, the classification of belong to different types of literature". [Conclusion] the results / experiments show that compared with the classification method has not been extended, classification The accuracy rate is increased by 5.5% to 8.4% respectively, which proves that the method can effectively improve the classification effect of automatic text classification.

【作者单位】：武汉大学信息管理学院;武汉大学信息资源研究中心;
【基金】：国家社会科学基金项目(15BTQ066)
【分类号】：TP391.1

【相似文献】