基于自动标引的《中分表》改造及测评研究  

Research on Revision and Evaluation of CCT Based on Automatic Indexing Data

作　　者：何琳[1] 何娟[1] 阎素兰[1] 

机构地区：[1]南京农业大学信息管理系

出　　处：《中国图书馆学报》2012年第6期81-88,共8页Journal of Library Science in China

摘　　要：本文探讨了基于自动标引的《中国分类主题词表》(简称《中分表》)改造的模式、结构以及关键技术。在原《中分表》分类体系的框架之上,收集标引经验库中分类标引和主题标引的双重标引数据及其他相关数据,应用支持度、置信度和相关度等筛选处理方法,最终得出分类号与关键词(串)的最佳对应关系组合。本文从收词量、相符度、专指度、标引深度、主题标引能力和分类标引能力6个方面详细地对改造后的《中分表》进行了测试,结果表明改造后的《中分表》在编制方式、类目设置、收词量、全面性和专指性等方面都具有一定优势。建议在《中分表》的更新改造中,尽量采用立体化的整体结构,保证完备的收词量,进行必要的分级化控制并扩大用户交互。This paper summarized research results on the automatic construction of CCT based on indexing data, which focus on construction mode, framework and main approaches. It collected all of the classification and subject inde- xing data from indexing database based on the framework of CCT to find the best corresponding of class number to descrip- tor by filtering processing method of confidence, support and degree of association. It also tested the construction results by vocabulary, association, specificity, indexing depth and external data in details. From the testing results, it proved that the revised CCT has some advantages on construction mode, class setting, vocabulary, comprehensive and specificity. At last, the paper suggested that it is very important for the construction of CCT to take a three-dimensional construction in achieving complete vocabulary, grade-based management and sufficient user interaction. 7 tabs. 9 refs.

关 键 词：《中国分类主题词表》 自动标引 自动构建 测评指标 

分 类 号：G254[文化科学—图书馆学]