PhD Studentship: Using Complex Networks and Machine Learning to Detect Situational Inauthentic Behaviour Online

University of Exeter

Project Title: Using Complex Networks and Machine Learning to Detect Situational Inauthentic Behaviour Online. PhD in Computer Science

Department of Computer Science, Streatham Campus, Devon, University of Exeter

The University of Exeter’s College of Engineering, Mathematics and Physical Sciences is inviting applications for a fully-funded PhD studentship to commence in September 2023 or as soon as possible thereafter.

The studentship will cover tuition fees plus an annual tax-free stipend of at least £16,062 for 3.5 years full-time, or pro rata for part-time study. 

Project Description:

Online social media have revolutionised how people consume information, and form and share their opinions. In a perfect world, this easy-cheap access to information would boost the global economy and the democratic processes. However, there is growing evidence showing the destructive power of malicious actors exploiting platforms’ vulnerabilities and human’s cognitive biases to nurture the current infodemic crisis.

The arms race between social media platforms and inauthentic accounts has already a long list of cycles. As usual, as technology develops increasing the capacity of detection, evaders also implement counter measurements creating more sophisticated deceivable agents. The literature already offers methodologies to detect different facets of this problem. For instance, (i) automated vs. organic, (ii) benign vs. malicious, (iii) coordinated vs. uncoordinated, and (iv) true vs. fake. These methodologies already empower us (the general public and the platforms) to combat fringe actors exploiting our vulnerabilities online. Unfortunately, none of these technologies is perfect, and bad actors often employ a hybrid approach mixing most of the above-mentioned facets.

Most of this technology relies on large amounts of historical data not only to build models for detection but also to perform “live” detection. Moreover, they tend to focus on labelling accounts rather than actions. For instance, Botometer, one of the most famous bot detection tools uses up to the last 200 tweets from an account in other to assess its “botiness”. Even though bot scores change over time, there is no association between an action or time leading to inauthentic behaviour. The project aims to evolve online inauthentic behaviour detection similarly as criminology has evolved from Lombroso’s classical criminology school of thought which focused on traits of criminals to environmental criminology.  Can we develop a methodology to detect situational misbehaviour online?

Essential Criteria:

  • Strong data analysis skills demonstrated by academic excellence or practical experiences
  • Strong mathematical or statistical background, with the ability to conduct modelling
  • Programming skills, with a very good knowledge of Python or R
  • Applicants for this studentship must have obtained, or be about to obtain, a First or Upper Second Class UK Honours degree, or the equivalent qualifications gained outside the UK, in an appropriate area of science or technology.

    If English is not your first language you will need to have achieved at least 6.0 in IELTS and no less than 6.0 in any section by the start of the project. 

    Alternative tests may be acceptable (see ).

    View or Apply
    To help us track our recruitment effort, please indicate in your cover/motivation letter where ( you saw this job posting.

    Job Location