All-in-one: Multi-task Learning for Rumour Verification
This work addresses the challenge of verifying rumours for social media and information analysis, but it is incremental as it builds on existing multi-task learning methods applied to this domain.
The paper tackles the problem of automatic rumour verification by proposing a multi-task learning approach that jointly trains rumour detection, tracking, and stance classification, improving performance over separate pipeline components.
Automatic resolution of rumours is a challenging task that can be broken down into smaller components that make up a pipeline, including rumour detection, rumour tracking and stance classification, leading to the final outcome of determining the veracity of a rumour. In previous work, these steps in the process of rumour verification have been developed as separate components where the output of one feeds into the next. We propose a multi-task learning approach that allows joint training of the main and auxiliary tasks, improving the performance of rumour verification. We examine the connection between the dataset properties and the outcomes of the multi-task learning models used.