Public defense Himanshu Verma, 28.08, 16:00, BC 420

Collaborators' interactions with the shared workspace are well assimilated in the established work practices. However, the role of these interactions in collaboration assessment is not thoroughly investigated. In this dissertation, we report on the quantitative analyses of the group interactions with shared content in a small-group collocated setting. We hypothesize that these interactions might contain latent social information (LSI), which can be used to model certain aspects of ongoing collaborative processes. To test our hypothesis, we conducted several user-studies to a) extract the different attributes of LSI, b) examine their relationship with visible collaborative processes, and c) investigate their role in the task performance. Our findings demonstrate that some attributes of LSI can be used to model the level of mutual understanding amongst collaborators. Additionally, these attributes are also strongly related to the task performance, and inform us about the division of labor strategies employed by the groups.