Reading China: Measuring Policy Change with Machine Learning

Abstract

While China’s industrialization process has long been a product of government direction until now there has been no quantitative measures of the Chinese government’s policy priorities over a long period of time. We fill this gap by devising the first of such measures, the Policy Change Index (PCI) of China, which runs from 1952 to the present. We use the full texts of the People’s Daily, the official newspaper of the Communist Party of China, as raw data to construct the PCI. Our method is based on LSTM networks (à la Hochreiter and Schmidhuber, 1997) and the CUSUM test (à la Page, 1954) to detect significant changes in the policy-importance of People’s Daily articles. This method allows us to infer the shift in the priorities of the Chinese government’s policies. The constructed PCI not only matches important policy changes that have taken place in China—such as the Great Leap Forward, the Cultural Revolution, and the economic reform program—but is also able to make short-term predictions about China’s future policy directions.

Presenters

Weifeng Zhong

Details

Presentation Type

Paper Presentation in a Themed Session

Theme

Media Theory

KEYWORDS

Propaganda, China, Media

Digital Media

This presenter hasn’t added media.
Request media and follow this presentation.