Research Interests

The overarching objective of my research is to leverage online data to understand and improve aspects of the real world, and revolves around the broad area of Computational Social Science. My research tends to be very interdisciplinary, mostly focusing on computational problems in Natural Language Processing, Social Media Mining and Social Computing, but also making use of domain knowledge from other disciplines such as Journalism, Psychology and Sociology to advance research in Social Data Science.

Linking online and offline events can be understood as a problem that can be tackled in two different directions, either by modelling the reflection of real world events in online platforms, or by assessing and mitigating the effect of online events in the real world. The list of research interests below is not exhaustive and keeps continually evolving, but it's meant to give a general idea of the kinds of research problems that I work on.

• Assessing and mitigating online harms. The Web is often being used for malicious purposes, which puts individuals and society at risk. To mitigate the effect of these malicious actions, there is a need to improve computational methods for tackling hate speech, misinformation, cyberattacks, biases and inequality, among others. Examples in this line of research include:
• Claim detection for fact-checking: having texts as input (social media timelines, TV debates, news articles) determine which of the sentences should be deemed claims worthy of fact-checking.
• Claim veracity assessment: linking claims with relevant evidence and determining the truth value of the claims.
• Abusive language detection: including detection of hate speech, cyberbullying.
• Detection and mitigation of biased language.
• Generalisability in social media. Social media data is diverse, posted by users with very different backgrounds, across different social media platforms, in different time periods, using different languages, etc. This poses important challenges for conducting generalisable research in social media and to develop classification models that can be broadly used and generalised.
• Analysis of emotions and opinions in social media. This includes tasks such as stance classification and sarcasm detection through social media.
• Computational journalism. Online platforms like social media are increasingly being used for newsgathering, which requires computational methods to assist both journalists and end users. Examples of research problems within this line of research include event detection, summarisation of news, recommender systems, fact-checking, finding information sources such as eyewitnesses or development of dashboards to assist with newsgathering.
• NLP for social media. For example, lexical normalisation of social media content, e.g. converting 'how r u m8?' into 'how are you mate?'.

If you are interested in doing a PhD or discussing potential collaborations within these or related areas of research, do get in touch.