Modelling argumentation in short text: A case of social media debate

The technological leaps of artificial intelligence (AI) and the rise machine learning have triggered significant progress in a plethora natural language processing (NLP) understanding tasks. One these tasks is argumentation mining which has received interest recent years regarded as key domain for future decision-making systems, behaviour modelling, problems. Until recently, modelling tasks, such computational schemes, were often tested controlled environments, persuasive essays, reducing unexpected behaviours that could occur real-life settings, like public debate on social media. Additionally, growing demand enhancing trust explainability AI services dictated design adoption schemes to increase confidence outcomes solutions. This paper attempts explore short text proposes novel framework detection under name Abstract Framework Argumentation Detection (AFAD). Moreover, different proof-of-concept implementations are provided examine applicability proposed very developing rule-based mechanism compare results with data-driven Eventually, combination deployed methods applied increasing correct predictions minority class an imbalanced dataset. findings suggest process provides solid grounds technical research while hybrid solutions potential be wide range NLP-related offering deeper human reasoning.