U-M researchers train tech tool to find relationship clues from written conversations

By | April 27, 2020

Social scientists have identified 10 dimensions to describe the nature of human relationships but little research has focused on how these concepts are expressed through written language, and what role they have in shaping social interactions. New research from U-M and Nokia Bell Labs has used crowdsourcing and a tech tool to detect how these characteristics are expressed in everyday language and how they shape social dynamics.

In particular, the researchers wanted to find out if conversations could provide insight into knowledge, wealth, education and mental illness, including suicide. Using natural language processing, the team predicted social dimensions, including the relationship between people, for example one of conflict or support, and the type of real-world communities they shape.