Machine learning processes relevant to the development of a Problem Structuring Platform

I argue for the contribution that two machine learning techniques can make to the operation of an online Group Support System (GSS) leading to the notion of a Problem Structuring Platform. Probabilistic topic modelling can be used to classify very large quantities of documents, as a form of Augmented Qualitative Analysis (AQA), with a view to building preliminary causal maps for use in a GSS. Large Language Models (LLMs) can be used to elicit general knowledge about problem contexts through the use of directed prompts. When these techniques are combined with the ideas of a scaffold component to help participants self-facilitate through a problem structuring process the possibility opens up for a Problem Structuring Platform with potential for achieving i) scale-up to large group workshops, ii) rapid deployment in fast-paced decision environments, and iii) arbitrating and moderating participant behaviours for ensuring procedural rationality and justice.<p></p>