“How Bullying is this Message?”: A Psychometric Thermometer for Bullying

Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers, pages 695–706, Osaka, Japan, December 11-17 2016. “How Bullying is this Message?”: A Psychometric Thermometer for Bullying Parma Nand, Rivindu Perera, Abhijeet Kasture School of Engineering, Computer and Mathematical Sciences Auckland University of Technology Auckland, New Zealand {pnand,rperera,akasture}@aut.ac.nz Abstract Cyberbullying statistics are shocking, the number of affected young people is increasing dramat- ically with the affordability of mobile technology devices combined with a growing number of social networks. This paper proposes a framework to analyse Tweets with the goal to identify cyberharassment in social networks as an important step to protect people from cyberbullying. The proposed framework incorporates latent or hidden variables with supervised learning to de- termine potential bullying cases resembling short blogging type texts such as Tweets. It uses the LIWC2007 - tool that translates Tweet messages into 67 numeric values, representing 67 word categories. The output vectors are then used as features for four different classifiers implemented in Weka. Tests on all four classifiers delivered encouraging predictive capability of Tweet mes- sages. Overall it was found that the use of numeric psychometric values outperformed the same algorithms using both filtered and unfiltered words as features. The best performing algorithms was Random Forest with an F1-value of 0.947 using psychometric features compared to a value of 0.847 for the same algorithm using words as features. 1 Introduction Harassment and bullying as common forms of unacceptable behaviour have been topics in psychological research for a long time. Harassment aims to discriminate people on the basis of race, sex, disability etc. and bullying aims to intimidate people. The use of the terms “cyberharassment” and “cyberbullying” is relatively new. Both terms involve the application of modern technologies to support negative human behaviour. The ubiquitousness and affordability of modern technology makes it easy for organisations and individuals to easily communicate via e-mails, chat rooms, message boards and social media which generates a huge amount of public and private data containing information reflecting the pulse of the society. Unfortunately, this accessibility of the technology has also created a forum for the expression of negative human emotions, some of which also gives rise to tragic outcomes such as suicides, self harm and mental illnesses. The aim of our research is to create a framework for detecting cyberharassment from textual communication data, so that systems can be implemented to catch and eliminate harassing texts before it is able to inflict harm. Bullying is a relationship issue that could result in significant emotional and psychological suffering with some even resulting in suicides (Boyd, 2007). Cyberbullying is even worse because it can follow the victim everywhere, happen at anytime and it is frequently anonymous. The number of victims in all age groups is growing worldwide. It is urgent to progress the research in this area in order to develop advanced methods for fast detection and prevention of cyberbullying cases. This involves analysis of a huge amount of social network textual data which is largely …

Content truncated to 500 words