Spam & coordinated bots
Recognise the campaign behind the messages.
How can AI identify coordinated spam and manipulation when generated messages avoid simple repetition?
This exploratory area examines coordinated abuse across messages, accounts and time. It asks how a defender can connect related activity even when language and presentation vary. The aim is to assess evidence of harmful behaviour while accounting for legitimate uses of generated content, different languages and uncertainty about an account's role.
Research areas
Combine semantic similarity with timing, account behaviour and interaction patterns to investigate coordinated activity.
Study detection across paraphrases, languages and changing campaign themes using controlled generated examples.
Investigate how evidence from several accounts or messages changes the assessment of an individual item.
Evaluate early detection, missed campaigns and false positives affecting legitimate communication.
Publications & related resources
Projects, publications and research materials will be added here.
Contribute to this research.
Explore student research, internships and collaboration with the lab.