In group decision making, consensus is crucial to ensure everyone agrees on a single decision. However, reaching a consensus can be challenging when individuals have diverse opinions and changing views. This article surveys recent research on opinion aggregation, which helps groups reach a collective decision by combining individual opinions. The authors discuss various methods, including direct democracy, where each member votes directly; indirect democracy, where representatives vote on behalf of their constituents; and deliberative polling, where participants engage in structured discussions before casting their ballots.
The article highlights the importance of moderation in consensus decision making, as moderators play a crucial role in proposing potential solutions that are likely to gain acceptance from all group members. The authors emphasize the need to account for factors such as agent inhomogeneity, uncertainty, time discounting effects, and dependencies between agents to ensure fair and accurate opinion aggregation. They also discuss the challenges of dealing with complex spaces of beliefs, where opinions are not easily quantifiable or reducible to a single number or ranking.
To address these challenges, the article introduces the concept of "opinion representation," which involves translating individual opinions into a common space for comparison and combination. The authors propose several techniques, such as vector space models, probabilistic opinion representations, and hybrid approaches that combine multiple methods. These techniques enable more accurate and efficient opinion aggregation, especially in complex decision-making contexts.
In conclusion, the article provides a comprehensive overview of recent advances in opinion aggregation, highlighting the importance of moderation, accounting for complexity, and using effective representation techniques to reach consensus in group decision making. By demystifying complex concepts with engaging metaphors and analogies, this summary aims to help readers grasp the essence of the article without oversimplifying the material.
Computer Science, Multiagent Systems