论文部分内容阅读
[目的/意义]针对英文文献主题自动标引任务中通用概念过度标引的问题,拟基于STKOS超级科技词表构建适用于理、工、农、医多个领域的英文通用概念表,用于有效过滤通用概念,降低标引噪音,提高文献主题自动标引的质量。[方法/过程]设计多维特征概念通用度计算算法,通过计算概念通用度自动遴选STKOS超级科技词表中的候选通用概念,结合专家审核,构建具有较好适用性的英文通用概念表。[结果/结论]初步建成的通用概念表概念总量2782个(含某一领域通用与多领域通用),其中典型通用概念366个,能较好体现概念通用性的强弱程度及领域差异。经应用测试,可有效降低通用概念对文献主题自动标引的噪音影响。
[Purpose / Significance] In view of the problem of over-generalization of common concepts in the automatic indexing task of English literature subject, it is proposed to construct a common English concept table applicable to various fields including science, industry, agriculture and medicine based on the STKOS super-technology vocabularies Effectively filter the common concepts, reduce the indexing noise, and improve the quality of the automatic topic indexing. [Method / Process] To design multi-dimension feature generalization algorithm to automatically select candidate universal concepts in STKOS super-technology vocabularies by calculating conceptual commonality, and to construct a common universal English concept table with good applicability through expert review. [Results / Conclusions] The preliminary concept list of universal concept table built in advance is 2782 (including general and multi-domain common in a certain field), among which 366 are typical and common concepts, which can well reflect the degree of conceptual universality and the field differences. The application of testing, can effectively reduce the common concept of automatic indexing of the subject literature noise.