Abstract: Studies suggest that people who engage in deliberate self-harm behavior often find easier to discuss self-harm-related thoughts and behaviors using social media, rather than in the physical world. Given that the users who post content related to self harm may be need some sort of intervention, we would like to be able to automatically identify this content automatically as a first step towards helping these users. Additionally, identifying such content online at a large scale allows us to analyse this behavior and to potential come to new insights about users who engage in self-harm. In this talk, I will present our efforts to understand self-harm content, and will describe automatic approaches to its detection. The first part of this work involves a comprehensive analysis of self-harm content on social media using different input cues. This analysis, the first of its kind on such a large scale, reveals a number of important findings. Then I will describe proposed frameworks that incorporate these findings to discover self-harm content under both supervised and unsupervised settings. The experimental results, on a large social media dataset from Flickr, demonstrate the effectiveness of the proposed frameworks, which constitute a first step towards automatically detecting self-harm content in social media.
by Neil O’Hare, Yahoo, USA.
Dr Neil O’Hare is a Senior Research Scientist at Yahoo, based in Sunnyvale, California. He received his PhD degree in Computer Science from Dublin City University in 2007. Prior to working in Yahoo Sunnyvale, he spent 3 years working in Yahoo Labs, Barcelona. His research interests are in the areas of multimedia information retrieval, multimedia classification based on content and metadata, user behavior analysis in multimedia search, managing personal media collections, geographic information retrieval, social network analysis, and computational aesthetics. He has published over 35 refereed conference and journal papers.