Deep reinforcement learning based ensemble model for rumor tracking

作者:

Highlights:

• By exploring plenty of basic models, we proposed an aggregated model named RL-ERT to solve the rumor tracking task.

• We propose a reinforcement learning based bagging algorithm to aggregate basic models into a macrocosm.

• We conduct experiments on public benchmark datasets, and the experimental results show the rationality and superiority of RL-ERT.

摘要

•By exploring plenty of basic models, we proposed an aggregated model named RL-ERT to solve the rumor tracking task.•We propose a reinforcement learning based bagging algorithm to aggregate basic models into a macrocosm.•We conduct experiments on public benchmark datasets, and the experimental results show the rationality and superiority of RL-ERT.

论文关键词:Rumor tracking,Natural language processing,Deep learning,Reinforcement learning

论文评审过程:Received 15 September 2020, Revised 23 December 2020, Accepted 14 March 2021, Available online 26 March 2021, Version of Record 27 September 2021.

论文官网地址:https://doi.org/10.1016/j.is.2021.101772