Early Detection of Rumors Based on BERT Model

Li Yuechen (), Qian Lingfei and Ma Jing
Additional contact information
Li Yuechen: Nanjing University of Aeronautics and Astronautics
Qian Lingfei: Nanjing University of Aeronautics and Astronautics
Ma Jing: Nanjing University of Aeronautics and Astronautics

A chapter in AI and Analytics for Public Health, 2022, pp 261-268 from Springer

Abstract: Abstract The continuous development of social media and the multiplication of information promote the generation and spread of Internet rumors. At present, social media mainly rely on manual audit to detect rumors, which will consume a lot of labor cost and time. However, the traditional rumor detection methods usually need to use a large number of comments and forwarding information generated by rumor propagation to calculate, which leads to the failure to detect rumors in time. This paper attempts to achieve early detection of rumors by using deep learning models of Bidirectional Encoder Representation from Transformers (BERT), only using the original text without using comment information and forwarding information. After training, the rumor detection model can be used to detect the rumor when there is no comment at the beginning of the rumor release, and can achieve good detection results in a few seconds. The experimental results show that better rumor detection results can be obtained by using the Bert model compared with other models without relying on other data. The first innovation of this paper is to try to solve the rumor detection problem by using the Bert model. Second, only using the rumor text data can detect the rumor in the early stage.

Keywords: Rumor detection; BERT model; Pre training; Deep learning (search for similar items in EconPapers)
Date: 2022
References: Add references at CitEc
Citations:

There are no downloads for this item, see the EconPapers FAQ for hints about obtaining it.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:prbchp:978-3-030-75166-1_18

Ordering information: This item can be ordered from
http://www.springer.com/9783030751661

DOI: 10.1007/978-3-030-75166-1_18

Access Statistics for this chapter

More chapters in Springer Proceedings in Business and Economics from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().