Context Aware Harassment Detection in Social Media [Overview]
Download as PPTX, PDF3 likes3,058 views
The document outlines a research initiative focused on context-aware harassment detection on social media, led by a team at the Ohio Center of Excellence in Knowledge-Enabled Computing. It highlights the impact of online harassment, existing shortcomings in detection methods, and the project’s objectives to develop advanced techniques using various data analysis approaches. The work aims to create operational tools that address significant social issues with real-world implications.
Context Aware Harassment Detection in Social Media [Overview]
- 1. Context-Aware Harassment Detection on Social Media Ohio Center of Excellence in Knowledge-Enabled Computing NSF Award#: CNS 1513721 Principal Investigators: Prof. Amit P. Sheth Co-Investigators: Prof. Valerie L. Shalin, Prof. Krishnaprasad Thirunarayan Other Faculty: Prof. Debra Steele-Johnson, Dr. Jack L. Dustin PhD Students: Lu Chen, Wenbo Wang, Monireh Ebrahimi, Kathleen Renee Wylds Master Student: Pranav Karan, Rajeshwari Kandakatla Collaboration with Beavercreek High School Ohio Center of Excellence in Knowledge-enabled Computing (Kno.e.sis) Department of Psychology
- 2. Kno.e.sis problem selection • Real-world impact through partnership/collaboration with domain scientists/end-users. • Often addresses important human/social/economic development challenges. • Real-world scale: at minimum, developing robust prototypes (open source tools/platforms) using real-world data and knowledge; often pursues multidisciplinary research. • End results usually involve operational applications/tools/systems, technology transfer, products, or services. • High-risk/high-impact; avoids incremental work.
- 3. Sample Challenges addressed Modeling Social Behavior for Healthcare Utilization in Depression – with Cornell Hospital Gender-Based Violence – With United Nations eDrugTrends: monitor cannabis and synthetic cannabinoid use Social and Physical Sensing Enabled Decision Support for Disaster Management and Response kHealth: Physical Cyber Social data for Personalized Digital Health Project Safe Neighborhood: Westwood Partnership to Prevent Juvenile Repeat Offenders Context-Aware Harassment Detection on Social Media
- 4. Harassment on Social Media 4 66% of internet users who have experienced online harassment said their most recent incident occurred on a social networking site or app. Social media harassment has a potentially serious negative social consequence, including emotional distress, social anxiety, clinical depression, and suicide. Online Harassment – Pew Research Center Facebook is the worst social network for bullying with 19-year-old BOYS the most common victims – Daily Mail
- 5. However … The social science theory of online harassment focuses mainly on the experience and types of harassment, but fails to explore the specific features of harassing exchanges that create the experience. Little effort addresses the automatic detection of harassment. Existing academic work on harassment detection usually applies machine learning for binary classification, relying on message content while ignoring message context. The social networking sites, Facebook, Twitter, and YouTube, have been accused of not taking effective action against harassment. 5
- 6. Research goals Develop comprehensive and reliable context-aware techniques (using machine learning, text mining, natural language processing, and social network analysis) to glean information about the people involved and their interconnected network of relationships, and to determine and evaluate potential harassment and harassers. 6
- 7. Overview of the proposed research 7
- 8. Ohio Center of Excellence in Knowledge-Enabled Computing More at: Project Web Site NSF Award #1513721 TWC SBE: Medium: Context-Aware Harassment Detection on Social Media Lot more at: http://knoesis.org
- 9. Kno.e.sis problem selection • Real-world impact through partnership/collaboration with domain scientists/end-users. • Often addresses important human/social/economic development challenges. • Real-world scale: at minimum, developing robust prototypes (open source tools/platforms) using real-world data and knowledge; often pursues multidisciplinary research. • End results usually involve operational applications/tools/systems, technology transfer, products, or services. • High-risk/high-impact; avoids incremental work.
- 10. Sample Challenges addressed Modeling Social Behavior for Healthcare Utilization in Depression – with Cornell Hospital Gender-Based Violence – With United Nations eDrugTrends: monitor cannabis and synthetic cannabinoid use Social and Physical Sensing Enabled Decision Support for Disaster Management and Response kHealth: Physical Cyber Social data for Personalized Digital Health Project Safe Neighborhood: Westwood Partnership to Prevent Juvenile Repeat Offenders Context-Aware Harassment Detection on Social Media