Presenter Information

Patrick FurmanFollow

Major

Computer Science

Anticipated Graduation Year

2023

Access Type

Open Access

Abstract

Cyberbullying is a widespread form of online harassment with serious negative consequences for victims. In a cyberbullying instance, participants can be classified as harassers, victims, or bystanders. Identifying the roles of participants in cyberbullying instances can facilitate more effective intervention in these instances. We propose a hierarchical attention network to automatically classify the roles of users in cyberbullying conversations on ASKfm, a social media platform where users can ask and answer questions anonymously. Our model combines word, sub-sentence, and sentence-level attention mechanisms to represent the structure of posts on the ASKfm platform and capture relevant features for classification.

Community Partners

Arizona State University

Faculty Mentors & Instructors

Dr. Yasin Silva, Computer Science Department

Supported By

Dr. Deborah Hall, Arizona State University

Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 License.

Share

COinS
 

Identifying Participant Roles in Cyberbullying Through Hierarchical Attention Networks

Cyberbullying is a widespread form of online harassment with serious negative consequences for victims. In a cyberbullying instance, participants can be classified as harassers, victims, or bystanders. Identifying the roles of participants in cyberbullying instances can facilitate more effective intervention in these instances. We propose a hierarchical attention network to automatically classify the roles of users in cyberbullying conversations on ASKfm, a social media platform where users can ask and answer questions anonymously. Our model combines word, sub-sentence, and sentence-level attention mechanisms to represent the structure of posts on the ASKfm platform and capture relevant features for classification.