Analyzing Chinese State-backed Twitter Influence Operations

This post uses Natural Language Processing (NLP) techniques and tweet and account information of over 23,000 distinct accounts captured by Twitters Public Safety team to identify and examine the effectiveness of China’s Twitter-based influence operations.


Overview

As US-China relations continue to fracture, Beijing backed representatives are turning towards the Trump administration’s favorite tool – Twitter – to shape public opinion. The sharp uptick in usage has signaled a coordinated effort to leverage western social media and promulgate state-views, often pertaining to sensitive domestic topics.

While Beijing’s foray into western-owned social media is relatively new, attempts to shape attitudes of young American minds is not. Confucius Institutes, international economic development initiatives and increasingly Tik-Tok, a burgeoning social media platform owned by Chinese company ByteDance, have been the main vehicles in disseminating views. However, a prevailing theme to these mechanisms is that each are Chinese owned, and Chinese funded, providing elevated degrees of authority and censorship.

The Twitter-based influence operation marks the first time the Chinese Communist Party has conducted a large-scale campaign on a foreign owned platform. But have they been successful?

This post uses Natural Language Processing (NLP) techniques and tweet and account information of over 23,000 distinct accounts captured by Twitters Public Safety team to identify and examine the effectiveness of China’s Twitter-based influence operations.

Data

Twitter has published data (User Accounts and Tweets) of state-backed entities flagged for attempting to manipulate Twitter – Through Influencing Elections And/or Other Civic Conversations​

Distinct Tweets: 348,607

Distinct Accounts: 23,749

Languages: 20+

Results

Screenshot (318)

Screenshot (320)