摘要
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。
Folklorists,like most practitioners in a field,understand the history of their discipline through a combination of their own reading and the consensus inherited from their graduate training and professional interactions. Disciplinary history,an effectively oral form of communication,codifies quickly. Highly contingent and random processesbecome widely understood as historically inevitable. In this preliminary report on a larger project examining the application of computational methodologies in the service of intellectual history,we explore the use of topic modeling as a way to understand the ebb and flow of topics and paradigms within a domain. Using JSTOR' s Data for Research application programming interface to access the contents of 6778 articles from there folklore studies journals( Journal of American Folklore,Western Folklore,Journal of Folklore Research),we used one form of topic modeling,Latent Dirichlet Allocation,to delineate 50 distinct topics drawn from 125 years of research publication. Of particular interest here was the legendary 'turn toward performance'in our field.
出处
《文化遗产》
CSSCI
2016年第5期69-82,共14页
Cultural Heritage