Introduction
Misinformation campaigns are rarely confined to a single platform. A false claim might start on a fringe blog, gain traction in a private Facebook group, and eventually go viral on Twitter. Traditional research often focuses on single-platform analysis, failing to capture the full trajectory of cross-platform information flows. This paper introduces Information Tracer, an automated framework designed to track, quantify, and visualize the spread of specific URLs across Twitter, Facebook, and Reddit.
System Architecture
The Information Tracer system operates through a four-stage pipeline designed for continuous monitoring and analysis. By combining multi-platform data collection with specialized metrics, the tool helps researchers and fact-checkers identify coordinated manipulation and understand how different types of fake news propagate across the web.
- Data Collection: The system queries the Twitter Academic API and CrowdTangle (for Facebook and Reddit) to retrieve posts containing a target URL. To manage API rate limits, the system focuses on "influential" posts that drive the majority of engagement.
- Data Aggregation: It computes impact indicators, including Total Interaction (likes, shares, etc.) and the Coefficient of Traffic Manipulation (CTM), which detects unnatural Twitter activity by analyzing the concentration of tweets among top users.
- Breakout Analysis: It calculates the "Breakout Scale" to measure how many platforms a URL has successfully penetrated.
- Visualization: A web-based dashboard provides macro-level scatter plots of impact indicators alongside micro-level network graphs and timelines.
Quantifying Cross-Platform Spread
A core innovation of this framework is the Breakout Scale, which categorizes the reach of information based on the number of platforms where it achieves significant popularity (e.g., exceeding 100 interactions).
As shown in the classification table, categories 0 through 3—measuring spread across zero to three or more social platforms—are currently operationalized within the system. This allows researchers to distinguish between localized content and content that has achieved broad cross-platform virality. Future iterations aim to incorporate offline media and celebrity amplification (categories 4-6).
Key Findings and Results
The authors evaluated the tool using two primary datasets: 1,427 fake news URLs from Google Fact Check and 501 COVID-19 related claims from the IFCN. Several key insights emerged:
- Platform Silos: Facebook often acts as a silo; over 90% of fake news originating as native Facebook content does not spread to other platforms. In contrast, 40% of fake news from Twitter, YouTube, or news domains broke out on at least two platforms.
- Topic-Platform Correlation: Specific narratives gravitate toward specific platforms. For instance, 5G conspiracy theories were predominantly found on YouTube, while election fraud claims were more prevalent on Twitter.
- Manipulation Detection: High interaction counts do not always indicate organic popularity. The CTM metric effectively identified instances where a small number of accounts artificially inflated a video's reach through repetitive spamming at verified users.
Conclusion and Future Work
Information Tracer provides a scalable solution for fact-checkers and researchers to monitor the complex ecosystem of online misinformation. While limited by API access to private groups and the lack of "impression" data (actual views), the tool's multi-platform approach offers a more comprehensive view of information spread than single-platform tools. Future work will focus on integrating direct web traffic logs and collaborating with platforms to access deeper engagement metrics.
Summary updated