Paper Type

Complete

Paper Number

PACIS2026-1523

Description

Older adults may interact with the same care robot in markedly different ways. Drawing on Socioemotional Selectivity Theory and the Selection–Optimization–Compensation model, long-term dialogue logs from 441 Hyodol users were analyzed. Nine conversational topic distributions were used for K-means clustering, identifying five profiles: Emotional Outpourers, Adaptive Companions, Routine Anchors, Relational Connectors, and Entertainment Seekers. These profiles reflected distinct interaction strategies rather than sociodemographic differences. Emotionally and relationally oriented clusters were associated with more favorable pre–post patterns in depression and loneliness, whereas routine- or entertainment-oriented clusters showed mixed patterns. The findings highlight the value of conversation-based personalization in care robot design.

Comments

14-Healthcare

Download

Share

COinS
 
Jul 5th, 12:00 AM

The Same Robot, Different Conversations: Behavioral Differences Among Older Adults Revealed Through Dialogue Data

Older adults may interact with the same care robot in markedly different ways. Drawing on Socioemotional Selectivity Theory and the Selection–Optimization–Compensation model, long-term dialogue logs from 441 Hyodol users were analyzed. Nine conversational topic distributions were used for K-means clustering, identifying five profiles: Emotional Outpourers, Adaptive Companions, Routine Anchors, Relational Connectors, and Entertainment Seekers. These profiles reflected distinct interaction strategies rather than sociodemographic differences. Emotionally and relationally oriented clusters were associated with more favorable pre–post patterns in depression and loneliness, whereas routine- or entertainment-oriented clusters showed mixed patterns. The findings highlight the value of conversation-based personalization in care robot design.