TY - GEN
T1 - Early Detection of Multilingual Troll Accounts on Twitter
AU - Miao, Lin
AU - Last, Mark
AU - Litvak, Marian
N1 - Funding Information:
I. INTRODUCTION Social media platforms allow easy and fast creation and exchange of user-generated content. Trolls are Internet users who attempt to manipulate opinion or sow discord by spreading disinformation, inflammatory and false information1. During the past several years, a large number of organizations utilize troll farms to distribute rumors, conspiracy, and speculation, in an attempt to manipulate public opinion on social media. As more and more countries start to weaponize opinion manipulation, social media has been flooded with troll accounts spreading fake news, propaganda, and misleading information. For example, Russia has been accused of using trolls on Twitter to engage in espionage, manipulation, and propaganda on social media2. According to research funded by the UK government, Russian internet trolls are spreading support for the invasion of Ukraine during the conflicts in 202234. At the end of January 2019, Twitter started to delete thousands of troll accounts that may attribute to the government of Russia, Iran, and Venezuela. This emphasizes the importance of detecting trolls to protect the public from inappropriate
Publisher Copyright:
© 2022 IEEE.
PY - 2022/1/1
Y1 - 2022/1/1
N2 - Internet troll farms have recently been employed as a powerful and prevailing weapon of information warfare. Even though different tactics may be utilized by different groups of state-sponsored trolls, our goal is to leverage identified troll data for revealing new emerging trolls generating multilingual content. In this work, we adopt a model agnostic meta-learning framework making use of previously released troll farm datasets for the early detection of newly-emerged troll accounts from identified or unidentified troll farms. The detection earliness of various models is evaluated using variable amounts of the earliest tweets from the tested accounts. To evaluate the proposed meta-model, we compare it to several classification models based on different types of account features. Our experiments demonstrate the effectiveness of the meta-model requiring as few as ten tweets to detect a troll account with an average accuracy of 94%.
AB - Internet troll farms have recently been employed as a powerful and prevailing weapon of information warfare. Even though different tactics may be utilized by different groups of state-sponsored trolls, our goal is to leverage identified troll data for revealing new emerging trolls generating multilingual content. In this work, we adopt a model agnostic meta-learning framework making use of previously released troll farm datasets for the early detection of newly-emerged troll accounts from identified or unidentified troll farms. The detection earliness of various models is evaluated using variable amounts of the earliest tweets from the tested accounts. To evaluate the proposed meta-model, we compare it to several classification models based on different types of account features. Our experiments demonstrate the effectiveness of the meta-model requiring as few as ten tweets to detect a troll account with an average accuracy of 94%.
KW - meta-learning
KW - multilingual classification
KW - troll account detection
KW - Twitter
UR - http://www.scopus.com/inward/record.url?scp=85151960035&partnerID=8YFLogxK
U2 - 10.1109/ASONAM55673.2022.10068705
DO - 10.1109/ASONAM55673.2022.10068705
M3 - Conference contribution
AN - SCOPUS:85151960035
T3 - Proceedings of the 2022 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, ASONAM 2022
SP - 378
EP - 382
BT - Proceedings of the 2022 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, ASONAM 2022
A2 - An, Jisun
A2 - Charalampos, Chelmis
A2 - Magdy, Walid
PB - Institute of Electrical and Electronics Engineers
T2 - 14th IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, ASONAM 2022
Y2 - 10 November 2022 through 13 November 2022
ER -