Supporting Learning & Collaboration in Online Discussion
Abstract
People engage in online discussion about various social issues, such as breaking news and longpending controversial issues. Different platforms afford different kinds of discussion, ranging from commenting on news sites (e.g., the New York Times) to exchanging opinions on dedicated discussion forums (e.g., Reddit) to discussing with friends and followers on social media (e.g., Twitter). While the amount of platforms for online discussion and the discussionsthemselves have been rapidly increasing, discussion quality and experience have suffered in various ways: prevalence of commentary over facts, little cost tocreating and sharing disinformation, curation and selective emphasis over authentic viewpoint, shortlived attention,access to only unchallenging and compatible views (i.e., echo chamber), and algorithmically generated news feed andrecommendations reinforcing the current viewpoint (i.e., filter bubble).We frame these problems with the lens of learning and collaboration . Most online discussion platforms do not provideusers with enough opportunities to learn to discuss better. Many users do not, cannot, or forget to reference facts,challenge their own views, or consider other viewpoints in constructing theirarguments, possibly because existing discussion platforms do not guide or train users to engage in healthy, informed,and constructive discussion. Furthermore, while competitive social interactions in discussion may lead to strongarguments, the interactions can become overly hostile and hamper people to engage in healthy discussion. It???s seldomthe case for users to collaborate by merging individual arguments into a stronger one or dividing work to construct anargument that covers a larger scope. As a result, most arguments in discussion tend to be isolated, narrow, andfragmented. We propose a novel online discussion platform that is designed with learning and collaboration asfirstclass objects. The main design goals of the system are: (1) to provide learning and collaboration opportunities tousers while engaging in online discussion, and (2) to generate a richlylabeled, wellstructured discussion representationas a result of the individual and collective discussion activities. Applying techniques in crowdsourcing, the systemprovides micro discussion tasks designed to help users learn and collaborate to construct strong arguments. Theresulting artifact generates a novel discussion representation, which serves as both a comprehensive summary of thediscussion and navigational and sensemaking tool for future users. The proposed approach contributes a noveltechnique and system for enhancing online discussion, by enabling scaffolded learning and collaboration for users ofonline discussion. At the same time, our approach contributes a novel application of crowdsourcing, by enablingcollective generation of a discussion artifact that provides additional value to other users. We expect to validate oursolution in both a controlled environment (e.g., lab study, paid crowdsourcing platform) and an inthewildenvironment (e.g., live deployment, inclass deployment). We also expect to find opportunities to work with existingdiscussion platforms to employ at least parts of our solution (e.g., micro discussion tasks, perspective taking) andintegrate them into these larger platforms. Another avenue for impact is from presenting and sharing the discussionrepresentation generated by our system. We believe this representation will yield unprecedented data about how adiscussion evolves over time by a crowd of users, such as : pertask outcomes, peruser contribution, andindividual/collective behaviors. We plan to open this data and resources to the public, so that interested researchersand developers can analyze, design, and build using them.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Sep 04, 2018
- Source ID
- N000141812834
Entities
People
- Juho Kim
Organizations
- KAIST
- Office of Naval Research
- United States Navy