This interactive plot illustrates networks of TikTok accounts that coordinated the sharing of a video with an identical description within a five-minute span. The networks were identified utilizing CooRTweet (https://github.com/nicolarighetti/CooRTweet), applied to TikTok Research API data, originating from a search for videos tagged with the #moskow hashtag. This serves as a proof of concept. Instruction: Click on any node to navigate to the corresponding user account in a new window.

198 clusters

199 clusters

Stories circulated on Facebook up to three months before the election day

Stories circulated on Facebook up to three months before the election day

Stories circulated on Facebook up to three months before the election day

Stories circulated on Facebook up to three months before the election day

The project utilizes a dataset comprising 38,099 links to stories shared on Facebook from 2017 to 2022. These stories were flagged as problematic by Meta's third-party fact-checkers. The primary objective is to identify and analyze emergent narratives, with a focus on examining their persistence and resilience over time and across various platforms. 2. Research Questions How to characterize the impacts of misinformation online? Does misinformation persist (on a story or narrative level) despite fact-checking and debunking efforts? How does misinformation adapt to fact checking? 3. Methodology To identify the topics, we employed LLMs at three distinct levels. Firstly, we used OpenAI's GPT-3.5-turbo-1106 to translate the titles and descriptions of all stories in English. These translated texts were then used to generate embeddings via the Mistral AI API. We clustered these embeddings into 87 distinct groups or topics using the k-means algorithm, based on their proximity. For the final step, we utilized GPT-4-1106-preview to label our clusters. The quality of these labels was assessed in the project's initial phase. Three separate groups of participants independently evaluated the accuracy of the labels, using a five-level scale ranging from “Very Accurate” to “Not Accurate at All”. This evaluation was conducted on a stratified sample of 400 stories. In the final stage, we mapped the stories onto a two-dimensional space. Each group then selected a subset of coherent topics for more in-depth analysis.

-

-

-

Clusters of false news obtained with OpenAI and Mistral AI embedding models

False stories country clusters

Made with CooRnet

Content to be embedded in http://coornet.org/abc.html

A visualization of shares timeline of President Conte's farewell post. Powered by CooRnet on CrowdTangle data.

Condivisioni pubbliche del post di Giuseppe Conte

Circulation of the main content of the case. The press release published by the RAI website introduces the weekly episode of Report (a popular Italian investigative journalist TV show). Google translated version at https://translate.google.com/translate?sl=auto&tl=en&u=https://www.rai.it/programmi/report/news/2021/01/Dalle-mail-dellEma-ritrovate-sul-dark-web-si-scopre-che-a-novembre-erano-emersi-problemi-nella-qualita-del-vaccino-Pfizer-integrita-dellmRna-inferiore-rispetto-alle-dosi-usate-nei-trials-5c533acd-bcfa-4f29-981e-2a70806cfda3.html.

-

-

with italian subtitles

Size represents the number of subscribers of each group of page that spread the video.

https://www.ilgazzettino.it/italia/cronaca_nera/vaccini_influenza_morti_lotti_fluad-719182.html

https://www.ilgazzettino.it/italia/cronaca_nera/vaccini_influenza_morti_lotti_fluad-719182.html

https://www.lastampa.it/topnews/economia-finanza/2019/04/07/news/un-italiano-su-tre-rivuole-la-lira-cresce-il-populismo-e-l-europa-vacilla-1.33693523

https://www.lastampa.it/topnews/economia-finanza/2019/04/07/news/un-italiano-su-tre-rivuole-la-lira-cresce-il-populismo-e-l-europa-vacilla-1.33693523

-

https://www.foxnews.com/politics/birx-says-government-is-classifying-all-deaths-of-patients-with-coronavirus-as-covid-19-deaths-regardless-of-cause

https://imgur.com/t/awesome/3864lgI

http://archive.vn/OJif3

https://en.wikipedia.org/wiki/Anthony_Fauci

http://archive.vn/YHd57