Copy the page URI to the clipboard
Anastasiou, Lucas
(2023).
DOI: https://doi.org/10.21954/ou.ro.00017252
Abstract
Deliberation is the process through which communities identify potential solutions for a problem and select the solution that most effectively meets their diverse requirements through dialogic communication. Online deliberation is implemented nowadays with means of social media and online discussion platforms; however, these media present significant challenges and issues that can be traced to inadequate support for Sensemaking processes and poor endorsement of the quality characteristics of deliberation.
This thesis investigates integrating computational argumentation methods in online deliberation platforms as an effective way to improve participants' perception of the quality of the deliberation process, their way of making sense of the overall process and producing healthier social dynamics.
For that, two computational artefacts are proposed: (i) a Synoptical summariser of long discussions and (ii) a Scientific Argument Recommender System (SciArgRecSys). The two artefacts are designed and developed with state-of-the-art methods (with the use of Large Language Models - LLMs) and evaluated intrinsically and extrinsically when deployed in a real live platform (BCause).
Through extensive evaluation, the positive effect of both artefacts is illustrated in human Sensemaking and essential quality characteristics of deliberation such as reciprocal Engagement, Mutual Understanding, and Social dynamics. In addition, it has been demonstrated that these interventions effectively reduce polarisation, the formation of sub-communities while significantly enhancing the quality of the discussion by making it more coherent and diverse.
CORE (COnnecting REpositories)
CORE (COnnecting REpositories)
