Introduction

More than three quarters of women worldwide have experienced online violence and abuse according to a report commissioned by the UN Broadband Commission in 2015. Amnesty International’s own research in the UK has revealed that women in leadership positions or from ethnic minority, religious, disability or LGBQT communities are the most susceptible targets.

To tackle this problem Data Study Group researchers were challenged to develop a tool that would enable quantifiable and real time monitoring of online violence against high profile individuals, and particularly women. This would enable Amnesty International to gather concrete evidence which could be used to hold social media platform owners such as Twitter to account and call on them to take action to protect their users.

Interview

Reham Al Tamime, PhD student in web sciences at the University of Southampton facilitated the challenge and worked with a team of 9 researchers with skills and expertise in natural language processing (NLP), network analysis, data analytics, neuroscience, computer science, computational biology, cognitive science and machine learning. 

Explaining the researchers’ approach to the challenge she says: "The team had a diverse range of backgrounds and skills and were able to contribute in multiple ways. Before embarking on the design of a training model we had to figure out how online abuse can be detected. The researchers were divided into three sub groups. The first group focused on finding good data annotations and removing bias to standardise what constitutes an abusive or non-abusive tweet. This was the first step to ensure data quality control. The second group worked on network analysis and visualisation. The team was able to set up a real time dashboard with the objective of making the behaviours of abuse more transparent. The third group worked on building and evaluating models to detect abusive tweets using NLP techniques. Overall these approaches complemented each other and none could have worked in isolation.”

Although limitations in quantity and quality of the data meant that the group were not able to develop a full working solution, Reham is positive that the work done during DSG is definitely a step in the right direction. She says: “We encountered some challenges because the quality and quantity of the data wasn’t sufficient. However, what we discovered as a result of the annotation experiment is that there was a lot of bias embedded in our own interpretations of what constitutes ‘abusive’ and we found that we couldn’t easily agree. We only tried to label 30 tweets so it was a very small data set, but our results were in line with other similar experiments that have been carried out by other researchers.”

She continues: “This is still useful for Amnesty and is something for them to take forward and build on. They are positive about the outcomes we came up with and are aware of the issues.”

"As a facilitator I could observe and learn about everyone’s work; I feel that I have learned so much"

Reham Al Tamime, Facilitator and PhD student at the University of Southampton

Reham was thrilled when her application to participate in DSG was accepted and even more pleased when she was selected to facilitate the challenge she had set her heart on. She says: "I was so happy, as this is something I haven’t done before. It’s a totally new experience for me and I feel that I have learned so much about data science and been given the chance to work with some really smart and amazing people. As a facilitator I could observe and learn about everyone’s work. Everyone has put in so much time and effort and accomplished a lot.”

She adds: “I think the team did a really great job and I thought they worked really well together. I was impressed by the sub-groups and how they made the effort to make it feel inclusive. They were easy to talk to and everyone was able to connect and contribute to the best of their ability.”

In response to the question about what was the best thing about the Data Study Group Reham replies: “Being able to do something completely different to what you do in your PhD and being able to experience and be part of a new project. I feel it was useful academically and personally too. It was definitely worthwhile and I would definitely do it again.”