Abstract: Social media have gained wide popularity as a medium that enable the users to extract information and breaking news. But all the information that spreading across this social media is accurate. One of the features that characterizes social media is the rapid emergence and spread of new information. This leads to the circulation of rumours. A rumour is defined as an unverified or unconfirmed statement or report circulating in a community. The rumour resolution process has been defined as a pipeline involving four sub-tasks: (1) rumour detection, determining whether a claim is worth verifying rather than the expression of an opinion; (2) rumour tracking, collecting sources and opinions on a rumour because it unfolds; (3) stance classification, determining the attitude of the sources or users towards the truthfulness of the rumour, and (4) rumour verification, as the ultimate step where the veracity value of the rumour is predicted. Here express the rumour resolution process as a multitask problem that needs to address a number of challenges, where the veracity classification task is the main task and therefore the remainder of the parts area unit auxiliary tasks that can be leveraged to boost the performance of the veracity classifier. Multitask learning refers to the joint training of multiple tasks, which has gained popularity recently for a range of tasks in Machine Learning and Natural Language Processing and has been connected in various diverse errands and machine learning structures. Its adequacy is essentially credited to learning shared portrayals of firmly related assignments. Here rumour verification is the main task and others are considered to be auxiliary tasks.
Keywords: Rumour Detection, Rumour Verification, Stance Classification, Multi-task Learning