用机器学习模型解释部分自由国家互联网与民主的关系
Explaining the Relationship between Internet and Democracy in Partly Free Countries Using Machine Learning Models
摘要：
以前的研究对民主和互联网之间的关系提供了各种解释。然而，这些研究大多集中在地区，特定的国家或专制政权。没有研究调查互联网在自由之家定义的部分自由国家的影响。此外，人们对网络审查对民主发展、停滞或衰落的影响知之甚少。利用国际电信联盟、自由之家和世界银行的数据库，并使用机器学习方法，这项研究揭示了互联网对部分自由国家民主化的影响。研究结果表明，互联网渗透和网络审查都有负面影响的民主分数和互联网的民主分数的影响是由网络审查的条件。此外，随机森林的结果表明，网络审查是最重要的变量，其次是治理指数和民主教育分数。通过对各种机器学习模型的比较，发现175树随机森林模型的预测精度最高，达到92%。此外，这项研究可能有助于“IT专业人员”看到他们的重要作用，不仅在技术领域，而且在社会方面的民主化和IT是多么接近社会科学。
英文摘要：
Previous studies have offered a variety of explanations on the relationship between democracy and the internet. However, most of these studies concentrate on regions, specific states or authoritarian regimes. No study has investigated the influence of the internet in partly free countries defined by the Freedom House. Moreover, very little is known about the effects of online censorship on the development, stagnation, or decline of democracy. Drawing upon the International Telecommunication Union, Freedom House, and World Bank databases and using machine learning methods, this study sheds new light on the effects of the internet on democratization in partly free countries. The findings suggest that internet penetration and online censorship both have a negative impact on democracy scores and the internet's effect on democracy scores is conditioned by online censorship. Moreover, results from random forest suggest that online censorship is the most important variable followed by governance index and education on democracy scores. The comparison of the various machine learning models reveals that the best predicting model is the 175-tree random forest model which has 92% accuracy. Also, this study might help "IT professionals" to see their important role not only in the technical fields but also in society in terms of democratization and how close IT is to social sciences.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
