摘要：

【目的】 为深入了解旱地农业这一特定领域内科技期刊发展和分布的特点,探索学科热点主题发展与期刊发展之间的联系,以旱地农业领域文献数据为基础,提出基于隐含狄利克雷分布(LDA)模型的科技期刊发展分析的方法和框架。【方法】 从LDA模型提取的旱地农业领域热点主题入手,自定义期刊热点主题追随度指数、领域重点期刊的遴选标准和分类标准,对不同类型的期刊进行热点主题追随度分析,并对学科领域期刊的分布和中国期刊发展概况进行多方位的数据提取和展示。【结果】 整个旱地农业领域研究的发展和重要期刊发展相互促进,不同类型期刊热点主题特点不一。发展速度快的重点期刊(包括顶级期刊、最快发展和较快发展重点期刊)中,所有旱地农业热点主题的相对增长率>0;非较快发展重点期刊和非重点期刊,其各个热点主题的增长率均<0。最快发展重点期刊紧随规模大且发展速度快的热点主题;顶级期刊和较快发展重点期刊则紧随旱地农业领域中发展速度相对快的方向。旱地农业领域发文期刊集中,77%以上论文出版于美国、荷兰、英国、瑞士、德国,仅4.3%的中国作者论文在本土出版。中国主要SCI期刊的期刊指数居中,刊出热点主题与全球一致性高。【结论】 应积极推动中国SCI论文和中国SCI期刊的良性互动,优先建设快速发展的科研领域期刊。期刊要抓住研究热点,且在不同发展阶段有不同侧重点。

关键词: 隐含狄利克雷分布模型, 主题聚类, 期刊发展分析, 重点期刊遴选, 热点主题追随度, 旱地农业

Abstract:

[Purposes] This paper aims to gain a deeper understanding of the development and distribution characteristics of scientific journals in the dryland agriculture field and explore the relationship between the development of subject hot topics and journal development. It proposes a method and framework for analyzing scientific journal development based on the latent Dirichlet allocation (LDA) model and relevant literature data. [Methods] According to the hot topics extracted by the LDA model in dryland agriculture, we customized the journal hot topic follow-up index, selection criteria, and classification criteria for key journals in this field. Meanwhile, we conducted hot topic follow-up analysis on different types of journals and performed multi-dimensional data extraction and display on the journal distribution and development overview of Chinese journals. [Findings] The research on the entire dry agriculture field and the development of important journals promote each other, and the characteristics of hot topics vary among different journal types. The relative growth rate of all hot topics in dryland agriculture in the key journals with fast development speed (including top journals, fastest developing key journals, and faster developing key journals) is more than 0, while that of each hot topic in non-fast developing key journals and non-key journals is less than 0. The fastest developing key journals closely follow the hot topics with large scale and fast development speed, and top journals and faster developing key journals are closely following the relatively fast direction in dryland agriculture. In dryland agriculture, there is journal publication concentration, among which over 77% of papers are published in the United States, the Netherlands, the United Kingdom, Switzerland, and Germany, and only 4.3% papers written by Chinese authors are published in China. China's major SCI journals rank in the middle of the index, and the hot topic publication has high global consistency. [Conclusions] Positive interaction between Chinese SCI papers and Chinese SCI journals should be advanced, and priority should be given to the construction of fast developing scientific journals. Additionally, journals should focus on research hotspots, with the research emphasis varying with the development stages.

Key words: Latent Dirichlet allocation model, Topic clustering, Journal development analysis, Key journal selection, Hot topic follow-up index, Dryland agriculture