Modeling Productive Social Positioning in Conversational Interactions
Effective exchange of information in doctor-patient conversations is critical for building trust and compliance with medical advice. In our past work, we have explored how patterns of information flow practices within doctor-patient interactions predict self-reported measures of trust from patients. Now we are building on extensive work modeling consensus-building practices in conversational interactions to work towards a conversational measure of shared decision making.
In particular, in this talk we probe into a specific quality of discussion referred to as Transactivity, both from a theoretical and a technical perspective. Transactivity is the extent to which a contribution articulates the reasoning of the speaker, that of an interlocutor, and the relation between them. In different contexts, and within very distinct theoretical frameworks, this construct has been associated with solidarity, influence, expertise transfer, and learning. Within the construct of Transactivity, the cognitive and social underpinnings are inextricably linked. From a theoretical perspective, extensive empirical work in support for collaborative group work will be described in order to motivate the application of this measure to doctor-patient interactions. Preliminary work on doctor-patient interactions will be reported. From a technical perspective, this talk will present a domain general deep learning approach to automatic measurement of Transactivity in human-human interactions and explore its application as a measure of interaction quality.