| Computer support for collaborative decision making has been a challenging task addressed by large research communities [22, 28, 32, 96, 102, 110, 117, 130, 159, 191]. Many solutions are provided by various tools and web-based social networks, and organizations struggle to find the best one to fit their needs [32, 110]. In this research, we address the issue of evaluating existing solutions from the perspective of how well they achieve elements needed for the support of collaborative decision making. This research is found at the confluence of the established research areas of discourse architecture, human-computer interfaces and artificial intelligence. We identify factors that are considered essential to the process of making decisions. Some of the most important of such factors consist of identifying the opposing arguments and evaluating their importance. We consider that these factors quantify how well participants learn and understand each other's positions, as well as how well they understand the addressed problem. We evaluate competing models for representing knowledge in debates regarding their impact on the aforementioned factors. To evaluate models, we keep the graphical user interface structure unchanged while modifying the allowed configurations of meta-data linking submitted comments (threading model). The identification and evaluation of the importance for arguments is based on interviews. All interviewed subjects are exposed to the same set of debates, but each of them encounters a different comment threading model. The debates are created by casting real televised debates into the evaluated threading model representation, a process that is improved by having it performed by multiple subjects.;The debates used in our research come from one area of applications, in particular, the area of theological argumentation. A question that arises is whether drawn conclusions do apply to other areas, such as politics and commerce. A set of surveys was used to compare the expectations of users from forums specializing in these different areas. It was found that expectations are different in two major regards, but a large number of features can still be shared between specialized forums.;In this dissertation, we detail our experiments and conclusions, explaining the principles used for preparing and conducting our surveys and interviews. The experiments validate the fact that quality differences between debate technologies can be quantified based on their final purpose. Three threading models are evaluated as a way to exemplify and test the concepts. This can lead to a principled and scientific improvement of electronic debate architectures, an important application of the computer technology.;This dissertation focuses on analyzing and comparing threading models for representing knowledge stemming from debates. Online networks, including Facebook and MySpace, provide an easy way to create circles of friends. However, they have not yet been studied from the perspective of how they improve group decision making. Large amounts of data and spam that can occur in decentralized social networks raise a challenge for debate user interfaces. The problem that emerges is to scientifically decide how well the debate mechanisms proposed in debate systems really offer the enhancements for which they were proposed. Our solution can be extended to address many of these problems. |
