Content-Based Features Predict Social Media Influence Operations

How easy is it to distinguish between industrialized information campaigns and organic social media activity? Our research finds that content-based features can help predict social media operations.

Abstract

We study how easy it is to distinguish influence operations from organic social media activity by assessing the performance of a platform-agnostic machine learning approach. Our method uses public activity to detect content that is part of coordinated influence operations based on human-interpretable features derived solely from content. We test this method on publicly available Twitter data on Chinese, Russian, and Venezuelan troll activity targeting the United States, as well as the Reddit dataset of Russian influence efforts. To assess how well content-based features distinguish these influence operations from random samples of general and political American users, we train and test classifiers on a monthly basis for each campaign across five prediction tasks. Content-based features perform well across period, country, platform, and prediction task. Industrialized production of influence campaign content leaves a distinctive signal in user-generated content that allows tracking of campaigns from month to month and across different accounts.

Background

The same features that make social media useful to activists — low barriers to entry, scalability, easy division of labor, and freedom to produce media targeted at any given country from almost anywhere in the world — also render it vulnerable to manipulation campaigns by well-resourced domestic and foreign actors. One well-covered example of these campaigns is the alleged effort by Russia’s Internet Research Agency (IRA) to shape American politics, for which it was indicted by the U.S. government in February 2018. A core scientific and policy question about this activity concerns how distinct industrialized information campaigns are from organic social media activity.

Study

To answer that question, we evaluate available Twitter data on Chinese, Russian, and Venezuelan troll activity targeting the United States, as well as the Reddit dataset of Russian influence efforts. First we design, train, and test a classifier algorithm on public activity to detect content that is part of coordinated influence operations based on content-based features such as timing, word count, and how material in a given post relates to other material in that period. Then, we use the classification framework to evaluate difficulty across five monthly tasks: 1. Distinguishes between influence-effort activity from normal activity; 2. Identifies social media posts from troll accounts; 3. Finds social media posts from troll accounts; 4. Detects activity across different platform data releases; and 5. Identifies social media posts from trolls using data from different platforms.

Results

We find that different campaigns have wide variance in our ability to detect them. The main results are: 1) Russia is the most sophisticated and difficult to track, whereas Venezuela is the easiest to track; 2) with relatively limited information about a campaign, it is still relatively easy to track its messaging into the next month; and 3) the further into the future we project, the worse our models perform, which demonstrates these campaigns are dynamic and evolve over time.