Common Sense Reasoning for Detection, Prevention, and Mitigation of Cyberbullying

作者: Karthik Dinakar , Birago Jones , Catherine Havasi , Henry Lieberman , Rosalind Picard

DOI: 10.1145/2362394.2362400

关键词:

摘要: Cyberbullying (harassment on social networks) is widely recognized as a serious problem, especially for adolescents. It much threat to the viability of online networks youth today spam once was email in early days Internet. Current work tackle this problem has involved and psychological studies its prevalence well negative effects While true solutions rest teaching have healthy personal relationships, few considered innovative design network software tool mitigating problem. Mitigating cyberbullying involves two key components: robust techniques effective detection reflective user interfaces that encourage users reflect upon their behavior choices.Spam filters been successful by applying statistical approaches like Bayesian hidden Markov models. They can, Google’s GMail, aggregate human judgments because sent nearly identically many people. Bullying more personalized, varied, contextual. In work, we present an approach bullying based state-of-the-art natural language processing common sense knowledge base, which permits recognition over broad spectrum topics everyday life. We analyze narrow range particular subject matter associated with (e.g. appearance, intelligence, racial ethnic slurs, acceptance, rejection), construct BullySpace, base encodes about situations. then perform joint reasoning wide life topics. messages using our novel AnalogySpace technique. also take into account analysis other factors. evaluate model real-world instances reported Formspring, networking website popular teenagers.On intervention side, explore set user-interaction paradigms goal promoting empathy among participants. propose “air traffic control”-like dashboard, alerts moderators large-scale outbreaks appear be escalating or spreading helps them prioritize current deluge complaints. For potential victims, provide educational material informs how cope situation, connects emotional support from others. A evaluation shows in-context, targeted, dynamic help during situations fosters end-user reflection promotes better coping strategies.

参考文章(53)
Janet Efstathiou, Introduction to Expert Systems McGraw-Hill, Inc.. ,(1991)
D. Schoen, The reflective practitioner ,(1983)
Helmut Prendinger, Alena Neviarouskaya, Mitsuru Ishizuka, Compositionality Principle in Recognition of Fine-Grained Emotions from Text international conference on weblogs and social media. ,(2009)
Justin W. Patchin, Cyberbullying Prevention and Response: Expert Perspectives Routledge, Taylor & Francis Group. ,(2011) , 10.4324/9780203818312
April Kontostathis, Lynne Edwards, Amanda Leatherman, Text Mining and Cybercrime Text Mining. pp. 149- 164 ,(2010) , 10.1002/9780470689646.CH8
Teresa Gonçalves, Paulo Quaresma, A Preliminary Approach to the Multilabel Classification Problem of Portuguese Juridical Documents portuguese conference on artificial intelligence. pp. 435- 444 ,(2003) , 10.1007/978-3-540-24580-3_50
Steven L. Salzberg, Alberto Segre, Programs for Machine Learning ,(1994)
Jacob L. Mey, Pragmatics: An Introduction ,(2001)
Simon Tong, Daphne Koller, Support Vector Machine Active Learning with Application sto Text Classification international conference on machine learning. pp. 999- 1006 ,(2000)
Catherine Havasi, Robert Speer, Amir Hussain, Erik Cambria, SenticNet: A Publicly Available Semantic Resource for Opinion Mining national conference on artificial intelligence. pp. 14- 18 ,(2010)