Machine learning-based consensus decision-making support for crowd-scale deliberation

Reason against the machine? Future directions for mass online deliberation

“…Similarly, Gürkan et al (2010) found that users required significant moderation support to input new ideas that fit the formalization of the argument map, and that user activity, such as "normal" conversations, moved outside of the deliberative platform. Yang et al (2021) also show that training algorithms to merely define what constitutes a positive, neutral and negative argument (to train automated facilitators that act upon that information), may seriously constrain the space and diversity of opinions. Furthermore, an assumption that unrefuted arguments are winning may unduly influence which arguments are of high quality.…”

“…Other tools allow crowd-based idea or argument harvesting (Fujita et al, 2017) or using the machine learning technique, case-based reasoning (CBR), to promote better idea generation, smooth discussion, reduce negative behavior and flaming and provide consensus-oriented guidance (Yang et al, 2019). In that way, the automated facilitation agent extracts semantic discussion structures, generates facilitation messages, and posts them to the discussion system, while the human facilitator can primarily focus on eliciting consensual decisions from participants (Yang et al, 2021).…”

“…They theorized that this would force participants to recognize flaws in their personal reasoning, motivate them to justify their demands, and ensure that they consider the viewpoints of others. Yang et al (2021) trained their algorithm on solutions human facilitators used in online discussion systems in the past. This means that the algorithm learns how to identify a given "problem" in a discussion (e.g., "a post or opinion that contains bad words or detouring information that may distract topic posts or main ideas") and then applies a "solution" (e.g.…”

“…However, obtaining high citizen engagement and maintaining high deliberative standards is particularly appealing now that there is a growing demand for citizen participation in urgent and important policy issues such as climate change or the energy transition (Schleifer and Diep, 2019). And even more so, because valuable and vital public ideas are inadequately reflected in current small-group deliberative practices (Yang et al, 2021). Indeed, the distribution of people's ideas is shown to have a long tail, thus requiring the participation of masses of people to ensure the diversity of ideas is adequately captured (Klein, 2012).…”

Towards Argument Mining for Social Good: A Survey

“…An example of this approach is represented in Deliberatorium 7 , an e-deliberation platform which has been extensively employed in many reference studies on the effect of digital innovation on deliberation (Klein, 2011). Another example of a digital deliberation platform which integrates argument maps and offers an option for moderation is COLAGREE (Yang et al, 2021;Ito, 2018). Among the studies testing the impact of such digital platforms on online deliberation, Spada et al (2015) tests the effect of Deliberatorium's argument maps on an online discussion among the supporters of the Italian Democratic party concerning the desired features of electoral law to be proposed by the party to the Parliament.…”