A research team led by Professor Juho Kim from the School of Computing has developed a system called SolutionChat that provides real-time moderator support during online chat discussions. The study was published in Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems.

As a result of their high accessibility and familiarity, online chats have become an increasingly popular medium for discussing important community matters. Examples include Comcast employees organizing a rally through the communication platform Slack, and Strong Towns, a non-profit organization, using the same platform to promote discussions about building financially resilient communities. Platforms like Slack are also ubiquitous in the academic field, utilized for project coordination or promoting student engagement. 

However, the synchronous nature of online chats can make it difficult for participants to keep track of the discussion, come to an optimal consensus, or catch up with missed conversations. Aiming to make discussions more structured and thereby more efficient, communities assign moderators who coordinate the flow and focus of the discourse by summarizing main points, giving an introduction to every new stage, and asking for supportive evidence.

While some existing platforms have been built to aid moderator work, few of them were designed to be applied to synchronous online chats. To get a better understanding of the challenges that go hand in hand with shaping structured discussions, the team conducted a formative study that revealed the limited awareness participants have of the discussion structure and a high burden in managing discussions in the midst of fast flow and high volumes of messages. Participants rely heavily on the moderator to remind them of what the current discussion stage is about, and many moderation messages show a recurring nature across different stages, e.g. the moderator has to send similar messages throughout discussions. To address these issues, the team introduced SolutionChat, a web-based chat interface that provides visualization of the overall structure of the discussion and featured opinions and suggests appropriate moderation messages to lower the task load. 

Through the agenda panel, a moderator can decide to move the discussion to the next stage, which is synchronized with all participants, and pin messages that include important opinions. By analyzing chat messages, current stage, featured opinions, and the elapsed time for each stage, SolutionChat recommends appropriate facilitation messages in real-time, which a moderator can simply click on to send to participants. Researchers believe that SolutionChat is the first system that offers both a summarization feature and a real-time messaging recommendation.