BPCPPDYSOE21 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

SOE: Fachverband Physik sozio-ökonomischer Systeme

SOE 2: Networks and Social Dynamics

SOE 2.4: Talk

Monday, March 22, 2021, 15:00–15:20, SOEa

Detection and Analysis of Fake News on Twitter — Zahra Ghadiri, Sima Hashemi, Milad Ranjbar, •Fakhteh Ghanbarnejad, and Sadegh Raeisi — Sharif University of Technology, Tehran, Iran

Fake news on social media has become a major problem that impacts many aspects of our lives. In this work, we try to combine ideas from complex systems and networks with techniques from natural language processing (NLP) to develop intelligent agents that can distinguish real and fake news. Our approach is based on the intuition that one of the more effective ways to detect fake news is to cross-check with reliable sources such as well-established news agencies. To this end, first we collect tweets from the Twitter accounts of official news agencies which are posted around the posting time of the target tweets. We use clustering algorithms to cluster tweets based on the topic and content. Next we identify the cluster that best matches the target tweet. Then we extract features from our tweets and train a classifier that based on the comparison with the corresponding cluster would identify fake tweets. This provides a NLP tool that enables us to check a posted tweet with news from news agencies or any other reliable source of information based on the content. We also build and investigate the evolution/dynamic trees of retweets. We analyze the topological features of the trees as well as the dynamical properties. We should note that there are challenges associated with the reconstruction of the network and dynamics of a tweet on Twitter that could potentially influence our results and conclusion.