Source-linked AI summary
Hoaxy: A Platform for Tracking Online Misinformation
Chengcheng Shao, Giovanni Luca Ciampaglia, Alessandro Flammini, Filippo Menczer
TL;DR
Online misinformation spreads at scale alongside fact-checking, creating difficult social-news dynamics to study. Hoaxy collects and tracks both through news-site and social-media data, and its preliminary analysis finds that fact checking follows misinformation by hours while involving a broader user base. The platform is designed as an observatory for researchers, journalists, and the public, but it does not itself determine content accuracy.
Problem
Misinformation and fact checking spread through complex, fast-changing social-media environments, making online news-sharing dynamics difficult to study.
Method
Hoaxy collects news stories from websites and their shared URLs from social media using crawlers, RSS, APIs, and stored structured data.
Results
Fact-checking content lagged misinformation by approximately 13 hours, while fact-checking tweets had more replies and quotes and involved a broader plurality of users.
Takeaways & Limitations
Hoaxy can help researchers, journalists, and the general public monitor and study the dynamics of online misinformation and related fact checking.
Takeaways & Limitations
Hoaxy does not perform fact checking and instead tracks shares from sources whose accuracy was determined independently.
Abstract
from arXiv · showhide
Massive amounts of misinformation have been observed to spread in uncontrolled fashion across social media. Examples include rumors, hoaxes, fake news, and conspiracy theories. At the same time, several journalistic organizations devote significant efforts to high-quality fact checking of online claims. The resulting information cascades contain instances of both accurate and inaccurate information, unfold over multiple time scales, and often reach audiences of considerable size. All these factors pose challenges for the study of the social dynamics of online news sharing. Here we introduce Hoaxy, a platform for the collection, detection, and analysis of online misinformation and its related fact-checking efforts. We discuss the design of the platform and present a preliminary analysis of a sample of public tweets containing both fake news and fact checking. We find that, in the aggregate, the sharing of fact-checking content typically lags that of misinformation by 10--20 hours. Moreover, fake news are dominated by very active users, while fact checking is a more grass-roots activity. With the increasing risks connected to massive online misinformation, social news observatories have the potential to help researchers, journalists, and the general public understand the dynamics of real and fake news sharing.
1. INTRODUCTION
Social media enable rapid, lightly overseen information sharing, allowing misinformation to spread widely while fact-checkers respond within the same attention-driven environment. Hoaxy addresses the resulting challenge by tracking misinformation and related fact checking for researchers, journalists, and the public.
- 65% of US adults access news through social media, where users can rebroadcast content without a single authority controlling distribution.This environment is vulnerable to the unintentional spread of false or inaccurate information.
- Misinformation spreads online in viral fashion and includes rumors, false news, hoaxes, and conspiracy theories.
- Fact-checking organizations produce timely verifications that are also consumed and broadcast through social media.Examples include Snopes.com, PolitiFact, and FactCheck.org.
- Homophily, polarization, algorithmic ranking, social bubbles, and fast news cycles complicate the study of social news-sharing dynamics.
- Hoaxy is a web platform designed to let researchers, journalists, and the general public monitor online misinformation and related fact checking.
- In a preliminary public-tweet analysis, fact checking lagged misinformation by approximately 13 hours, while fact-checking information involved a broader plurality of users.Misinformation was produced in much larger quantity than fact-checking content.
2. RELATED WORK
Prior systems addressed social-media abuse, content credibility, and rumor exploration through varied automated or semi-automated approaches. Hoaxy differs by emphasizing automatic tracking of news sharing rather than performing fact checking itself.
- Existing systems target political abuse, content credibility, or rumor detection using network analysis, content features, metadata, or interactive exploration.
- RumorLens, TwitterTrails, and FactWatcher require users to input a specific rumor instead of automatically monitoring the social-media stream.
- Hoaxy does not perform fact checking; it tracks news shares from sources whose accuracy has been independently determined.
3. SYSTEM ARCHITECTURE
Hoaxy combines monitored news websites and social-media streams to collect, organize, and track misinformation and fact-checking activity. Its architecture uses crawlers, RSS, APIs, and a database as the current foundation for future analysis.
- Hoaxy’s main objective is a uniform, extensible platform for collecting and tracking misinformation and fact checking.The currently implemented system focuses on the Monitors component.
- The system collects story origins and evolution from news websites, and shared story URLs from social media.
- Web scraping, syndication, and social-network APIs collect data, including real-time Twitter news-sharing through the streaming API.Tweets commonly share news through links to web articles, enabling domain-focused filtering.
- RSS supplies a unified protocol for collecting updates, while deep and light crawls acquire existing stories and subsequent feed changes.Collected structured data are stored in a database for convenient retrieval and planned interactive-dashboard analysis.
4. PRELIMINARY ANALYSIS
A preliminary analysis of public tweets compares misinformation with fact-checking across aggregate volume, timing, user activity, and sharing behavior. Misinformation is more prevalent, precedes fact-checking, and is promoted by a concentrated group of highly active accounts, whereas fact-checking is more distributed.
- Tweet Volume: Misinformation tweets exceed fact-checking tweets by approximately one order of magnitude in aggregate volume.The comparison reflects coverage of 71 fake-news domains versus six fact-checking websites.
- Tweet Volume: Approximately 13 hours separates misinformation sharing from subsequent fact-checking in the aggregate cross-correlation analysis.The analysis uses lags from −48 to +48 hours and a centered 24-hour moving average; larger-lag correlations are not excluded.
- Tweet Volume: Aligned spikes and decay patterns in two URL-matched examples suggest similar temporal activity despite low data volumes.The examples cover a Syrian-conflict story and the Alan Rickman death rumor, with 15 fake-news matches and two fact-checking matches for the latter.
- User Activity and URL Popularity: User activity and URL popularity display heavy-tailed, power-law distributions, with estimated decay exponents of 2.7 for activity and 2.5 for both popularity measures.URL-popularity fits use tails with n ≥200 and p ≥200, respectively.
- User Activity and URL Popularity: Fact-checking tweets contain more replies and quotes (> 20%) than misinformation tweets (≈10%), suggesting that fact-checking is more conversational.Original tweets and retweets comprise 80–90% overall, while quotes and replies comprise 10–20%.
- User Activity and URL Popularity: Rumor-mongering is dominated by few very active accounts, whereas fact-checking propagation is more distributed and grass-roots.Among top fake-news spreaders, the original-to-retweet ratio is much higher because they post many original promotional messages and retweet less.
5. CONCLUSIONS & FUTURE WORK
Hoaxy’s preliminary results point to an interplay between misinformation promoted by a few active accounts and grassroots fact-checking responses that arrive later. Future work will examine whether active spreaders are social bots and how timing varies across news types.
- Few very active accounts promote fake news, whereas fact-checking spreads through more grassroots responses several hours later.
- Hoaxy is intended to support the study of online misinformation dynamics by automatically tracking information sharing.
- Future studies will investigate whether active fake-news spreaders are social bots.
- Future analysis will expand to more news stories and examine how misinformation–fact-checking lag varies by news type.