AAAI Publications, Twelfth International AAAI Conference on Web and Social Media

Font Size: 
Real-World Witness Detection in Social Media via Hybrid Crowdsensing
Stefano Cresci, Andrea Cimino, Marco Avvenuti, Maurizio Tesconi, Felice Dell'Orletta

Last modified: 2018-06-15


The task of witness detection in social media is crucial for many practical applications, including rumor debunking, emergency management, and public opinion mining. Yet to date, it has been approached in an approximated way. We propose a method for addressing witness detection in a strict and realistic fashion. By employing hybrid crowdsensing over Twitter, we contact real-life witnesses and use their reactions to build a strong ground-truth, thus avoiding a manual, subjective annotation of the dataset. Using this dataset, we develop a witness detection system based on a machine learning classifier using a wide set of linguistic features and metadata associated with the tweets.


witness detection; crowdsensing; machine learning; Twitter

Full Text: PDF