RIDIR: Collaborative Research: Computational and Historical Resources on Nations and Organizations for the Social Sciences (CHRONOS)

该项目将收集、处理和分析数百万份有关国际关系的美国政府记录，开发探索这些记录的工具，并通过应用程序编程接口将所有这些记录放在一个单一的网站上。该项目将展示计算技术如何在一系列重大公共利益领域帮助定性和定量社会科学研究，扩展有关恐怖主义、情报、国际贸易和援助的知识。在其更广泛的影响中，它将改善多学科研究和教学的基础设施，并使公民、记者和民间社会组织能够更好地获得有关国际关系的信息。参与的学生研究人员将学习——并为——在“大数据”时代保持政府透明和问责的实用方法做出贡献。数字化或“天生数字化”文档的指数级增长将使这些方法在未来几年变得越来越重要。为了应对这一挑战，该团队将利用自然语言处理领域的新成果，定制将文本转换为数据的现有工具，并开发用于历史文档的新工具。他们将采用命名实体识别技术来提取人物、国家和组织的名字，使用户能够跟踪大型数字档案中提及的绝对频率和相对频率。通过主题建模，他们将总结主题内容，并展示最重要的主题如何随时间变化。而社会网络的提取将使他们能够揭示日常生活中形成政策的非正式关系。通过将所有这些量化的解密数据汇集到一个平台上，该项目将推动政治科学家以及通信和社交网络学者的工作，他们试图了解组织内部和组织之间的议程设置、权力和影响。该平台将使研究人员能够测试国际关系理论中的基本问题，例如政策制定者是否通常以现实主义者坚持的“国家安全”或自由主义假设的“国际规范”来谈论彼此。它将允许用户在不同层次的分析之间切换，从整个档案的汇总视图，到过滤后的元数据子集，再到产生单个数据点的句子中的特定单词。它将把定量和定性方法结合在一起，用于国际关系研究，并使两者更加透明、严谨和可复制。

This project will collect, process, and analyze millions of U.S. government records concerning international relations, develop tools to explore these records, and make all of them available on a single website with an Application Programming Interface. The project will demonstrate how computational techniques can aid both qualitative and quantitative social science research on a range of areas of major public interest, expanding knowledge about terrorism, intelligence, international trade and aid. Among its broader impacts, it will improve the infrastructure available for multidisciplinary research and teaching, and also give citizens, journalists, and civil society organizations much better access to information about international relations. Participating student researchers will both learn about -- and contribute to -- practical methods to keep government transparent and accountable in the age of "big data."The exponential growth in digitized or "born digital" documents will make such methods increasingly important in years to come. To meet the challenge, the team will draw on new work in Natural Language Processing, customize existing tools that turn text into data, and develop new tools for use with historical documents. They will employ Named Entity Recognition techniques to extract names of people, countries, and organizations, enabling users to track the absolute and relative frequency of mentions in large digital archives. Through Topic Modeling, they will summarize the thematic content and show how the most important subjects change over time. And Social Network extraction will allow them to reveal the informal relationships that shape policy from day-to-day. By bringing together all of this quantitative declassified data in a single platform, the project will advance work by political scientists as well as scholars of communications and social networks who seek to understand agenda-setting, power, and influence within and between organizations. The platform will enable researchers to test fundamental questions in IR theory, such as whether policymakers generally speak to one another in terms of "state security," as realists insist, or "international norms," as liberalism assumes. It will allow users to shift between different levels of analysis, from an aggregate view of whole archives, to filtered subsets of metadata, to the specific words in one sentence that produced a single data point. It will bring together quantitative and qualitative approaches to research in international relations, and make both more transparent, rigorous, and replicable.