Fake News Detection
The spread of rumour related to an event on a social media platform affects the dissemination of true information. Identifying rumour from multiple post becomes a challenging task as the rumour is similar to actual information. Rumour detection can be seen as a text classification problem.
PHEME dataset for Rumour Detection and Veracity Classification -
Kochkina, Elena; Liakata, Maria; Zubiaga, Arkaitz (2018): PHEME dataset for Rumour Detection and Veracity Classification. figshare. Dataset. https://doi.org/10.6084/m9.figshare.6392078.v1
Further based on the 9 events in the dataset more data on rumours was collected from Twitter.
Classified the tweets following a rumour by its stance into one of supporting, denying, questioning or commenting. A tweet agreeing with the source tweet is labeled as supporting, a tweet disagreeing with the source tweet is labeled as denying and so on. Also, a tweet disagreeing to a denying tweet is labeled as suporting.
A probablistic framework such as Hawkes Processes (HP) has been used in the modelling.