论文部分内容阅读
像大部分学者一样,民俗学者通过两方面的结合来了解自己的学科历史,一方面是自身的阅读,另一方面是源自研究型训练和专业互动的共识。民俗学科历史被快捷梳理,高度偶然随机的过程逐渐被普遍认可为历史的必然性。本文是一个检验计算机方法在思想史研究中应用大型项目的初步报告,我们对运用主题建模方式去理解某一领域中的主题及范式的兴衰进行了探究。我们使用期刊数据库的研究数据应用程序接口访问了源自三份民俗研究期刊(美国民俗学刊,西部民俗,民俗研究)的6778篇文章的内容,运用“潜在狄利克雷分布”(Latent Dirichlet Allocation)主题建模形式,描述了125年间的学术作品的50种不同主题。其中最有趣的就是在我们学科中著名的“转向表演研究”现象。
Like most scholars, folklorists learn about their subject history through a combination of two aspects: one is reading on the one hand, and a consensus derived from research-based training and professional interaction on the other. History of folk custom is quickly combed, a highly random process has gradually been generally recognized as the historical necessity. This article is a preliminary report examining the use of computer methods in the study of the history of ideas in large-scale projects. We explored the use of topic modeling to understand the themes and paradigms in a particular field. Using the Research Data API of the Journal Database, we interviewed 6778 articles from three folklore journals (American Folklore, Western Folklore, Folklore Research) using the “latent Dirichlet distribution” Dirichlet Allocation), a theme modeling style that describes 50 different topics for 125 years of academic work. One of the most interesting is the well-known “turn to show research” phenomenon in our discipline.