Presentation | 2022-11-24 A Comprehensive System for Social Robot Detection Using RoBERTa Classifier and Random Forest Regressor with Similarity Analysis Yeyang Chen, Mondher Bouazizi, Tomoaki Ohtsuki, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | As one of the most popular social media platforms, Twitter has skyrocketed over the past few years. At the same time, the number of social robots on Twitter has also increased significantly. To avoid being detected, social robots not only disguise themselves as normal users in the profile but also post tweets like real users to manipulate public opinions and affect the normal communication of users. Despite extensive research efforts, bot accounts on Twitter are still evolving to evade detection. Extracting reliable features is considered an effective method, but given the complexity of social robot behaviors, characteristics of the social robots extracted from a single perspective cannot accurately identify different types of robots. In our research, we propose a comprehensive system for social robot detection. The system comprehensively considers the information of different feature groups including profile-level features and tweet-level features. In addition, we propose the similarity of tweets and the predictions of the RoBERTa (Robustly Optimized Bidirectional Encoder Representations from Transformers Pretraining Approach) classifier as two new feature groups. We used the RoBERTa classifier to analyze the text of the tweets and a random forest regressor to analyze other features. We conduct experiments using the Twibot-20 dataset, and the results show that the accuracy of our system is up to 0.8588, which is higher than all other baseline methods. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Social MediaBot DetectionSimilarityVoting ClassifierRandom Forest RegressorRoBERTa |
Paper # | SeMI2022-51 |
Date of Issue | 2022-11-17 (SeMI) |
Conference Information | |
Committee | SRW / SeMI / CNR |
---|---|
Conference Date | 2022/11/24(2days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Epinard Nasu |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | IoT Workshop |
Chair | Hanako Noda(Anritsu) / Koji Yamamoto(Kyoto Univ.) / Masayuki Kanbara(NAIST) |
Vice Chair | Keiichi Mizutani(Kyoto Univ.) / Kentaro Saito(Tokyo Denki Univ.) / Hirokazu Sawada(NICT) / Kazuya Monden(Hitachi) / Yasunori Owada(NICT) / Shunsuke Saruwatari(Osaka Univ.) / Yuri Nishikawa(AIST) |
Secretary | Keiichi Mizutani(KUT) / Kentaro Saito(Niigata Univ.) / Hirokazu Sawada(NTT DOCOMO) / Kazuya Monden(Tokyo Univ. of Agri. and Tech.) / Yasunori Owada(Osaka Univ.) / Shunsuke Saruwatari(Toshiba) / Yuri Nishikawa(Nagoya Sangyo Univ.) |
Assistant | Maki Arai(Nihon Univ.) / Yuichi Masuda(Univ. of Tokyo) / Yuki Matsuda(NAIST) / Akihito Taya(Aoyama Gakuin Univ.) / Takeshi Hirai(Osaka Univ.) / Yuta Hoshi(NHK) / Junji Yamato(Kogakuin Univ.) |
Paper Information | |
Registration To | Technical Committee on Short Range Wireless Communications / Technical Committee on Sensor Network and Mobile Intelligence / Technical Committee on Cloud Network Robotics |
---|---|
Language | ENG |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | A Comprehensive System for Social Robot Detection Using RoBERTa Classifier and Random Forest Regressor with Similarity Analysis |
Sub Title (in English) | |
Keyword(1) | Social MediaBot DetectionSimilarityVoting ClassifierRandom Forest RegressorRoBERTa |
1st Author's Name | Yeyang Chen |
1st Author's Affiliation | Keio University(Keio Univ.) |
2nd Author's Name | Mondher Bouazizi |
2nd Author's Affiliation | Keio University(Keio Univ.) |
3rd Author's Name | Tomoaki Ohtsuki |
3rd Author's Affiliation | Keio University(Keio Univ.) |
Date | 2022-11-24 |
Paper # | SeMI2022-51 |
Volume (vol) | vol.122 |
Number (no) | SeMI-278 |
Page | pp.pp.12-17(SeMI), |
#Pages | 6 |
Date of Issue | 2022-11-17 (SeMI) |