Online deliberation is a relatively new form of online discussion that seeks to inform, marshal evidence, and provide structure for policy decisions. This purpose brings with it a need to maintain focus, support development and communication of competing positions, and allow varying formulations of the issues. As a key element of 21st century governance, successful deliberation environments can bring great benefits, such as broadening participation and tapping more expert knowledge. For online deliberation to be successful, participants must be motivated to contribute, and they must be able to identify where they can make their most valuable contribution. Beyond that, the contributions must subsequently be organized in such a way as to make the collection more usable by policy makers and their staff. The solution to both of these challenges is to provide a personalized navigable structure that reveals the shape and foci of the debates in a way that reflects the goals, expertise, and interests of users.
Prior work in the area of online communities suggests that people are willing to contribute more when they are made aware of needs that make their skills relevant. However, for realistic problem spaces, it is difficult to find where one could best make a contribution, understand how various contributions fit together, or understand the contingencies between needs and contributions. Making it easier to navigate through content and identify participation opportunities matching a user’s interests and abilities would likely improve both participation and quality of deliberation. However, existing approaches to navigable structure, such as forced argument mapping, are overly constraining, cumbersome, or difficult to learn how to use effectively. Neither do they handle the social aspects of deliberation, nor balance the attraction of joining like-minded peers with mitigating information bubbles. None has been successful in sustaining meaningful and effective deliberation. Furthermore, the issue of how to reorganize the collected contributions into a form that is then usable by policy makers is still unsolved in practice.
We propose to adapt a combination of crowdsourcing, machine learning, and social network analysis techniques to the problem of structuring, organizing, and providing entry points into online deliberations. From a design standpoint, this proposal seeks to provide a personalized view of the environment that will allow participants to see how their goals and interests match current themes and find groups of people or related sets of contributions that would be valuable for them to be aware of in meeting their participation goals. From a technical standpoint, this proposal seeks to integrate insights from sociolinguistics with state-of-the-art latent variable modeling techniques from the field of language technologies to extend prior work in the areas of perspective and stance modeling in order to identify the necessary structure in textual data to enable to nuanced form of personalized extraction, summarization, and presentation required. The system will also learn from users, as they add or remove links, for example, bringing crowdsourced human intelligence to the problem of structuring.
Intellectual Merit: We will systematically explore the effects of design decisions on participation, navigability, the nature of the deliberation. The design principles, interaction paradigms, and algorithms developed in this work will lead to more effective online deliberation systems. Algorithms and paradigms that help people find contribution opportunities may also apply more generally in other online communities, to increase the effective use of human resources to meet community needs. Algorithms that help find and form connections between topics will push existing computational tools for clustering, topic modeling to new levels of usefulness, through greater ability to interact with and learn from human users.
Broader Impacts: Online deliberation has potential to enhance democracy and give the marketplace of ideas broader reach and greater comprehensibility. The work will be employ a user-centered development methodology, including use in support of policy-oriented courses, experiments to more precisely gauge the effects of the manipulations on policy positions and deliberations, and at scale to support actual deliberation on real issues of concern and to evaluate the approach. Prototype deployments in an educational setting that effectively engage students in important policy debates will have immediate educational benefits, as well as long-term benefits as graduates become active in policymaking.