From social media narratives to experiential networks: A Large Language Model (LLM) multi-agent framework for mapping the integrated mobility-theme-affect network of Citywalk

Citywalk marks the rise of experiential travel that connects cultural, commercial, neighborhood, and natural places through walking sequences, in which route themes and affective appraisals are central to how journeys are constructed and narrated. Existing mobility research, which relies largely on structured data such as GPS trajectories, often struggles to capture the nuances of this travel behavior as documented in social media narratives. This study introduces an interdisciplinary framework integrating large language model (LLM) agents and network science to transform unstructured social media narratives into travel-behavior insights. Using Chain-of-Thought prompting, specialized LLM agents extract sequential locations, classify behavioral themes, and interpret affective states through the Pleasure-Arousal-Dominance (PAD) framework. These outputs form a directed, weighted mobility-theme-affect network representing narrated spatial pathways and environmental appraisals. Applying this framework to a Xiaohongshu (RedNote) dataset for Xiamen reveals a clustered experiential topology concentrated in the historic urban core. The results identify a robust core-periphery structure anchored by several prominent hubs that organize the city’s experiential landscape. Multidimensional analysis of these hubs further shows that a location’s network centrality does not always correspond to its social media mention frequency. Community detection further reveals that the citywide network comprises geographically coherent walking systems, while shared nodes assume different thematic and affective roles across route contexts. By transforming subjective narratives into quantifiable network structures, this approach advances the conceptualization and measurement of travel behavior and provides affect-informed insights for active mobility management and urban governance.

Authors

Institutions

Publication Details

Journal
Travel Behaviour and Society
Published
2026-11-14
DOI
https://doi.org/10.1016/j.tbs.2026.101439
Primary Topic
Human Mobility and Location-Based Analysis
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

From social media narratives to experiential networks: A Large Language Model (LLM) multi-agent framework for mapping the integrated mobility-theme-affect network of Citywalk

Yuxin Liu, Weipeng Deng, Zhang Chengbo, Yuxuan Cai et al.
Travel Behaviour and Society
Human Mobility and Location-Based Analysis
article

From social media narratives to experiential networks: A Large Language Model (LLM) multi-agent framework for mapping the integrated mobility-theme-affect network of Citywalk

Yuxin Liu, Weipeng Deng, Zhang Chengbo, Yuxuan Cai, Dandi Zhu
article en

Abstract

Citywalk marks the rise of experiential travel that connects cultural, commercial, neighborhood, and natural places through walking sequences, in which route themes and affective appraisals are central to how journeys are constructed and narrated. Existing mobility research, which relies largely on structured data such as GPS trajectories, often struggles to capture the nuances of this travel behavior as documented in social media narratives. This study introduces an interdisciplinary framework integrating large language model (LLM) agents and network science to transform unstructured social media narratives into travel-behavior insights. Using Chain-of-Thought prompting, specialized LLM agents extract sequential locations, classify behavioral themes, and interpret affective states through the Pleasure-Arousal-Dominance (PAD) framework. These outputs form a directed, weighted mobility-theme-affect network representing narrated spatial pathways and environmental appraisals. Applying this framework to a Xiaohongshu (RedNote) dataset for Xiamen reveals a clustered experiential topology concentrated in the historic urban core. The results identify a robust core-periphery structure anchored by several prominent hubs that organize the city’s experiential landscape. Multidimensional analysis of these hubs further shows that a location’s network centrality does not always correspond to its social media mention frequency. Community detection further reveals that the citywide network comprises geographically coherent walking systems, while shared nodes assume different thematic and affective roles across route contexts. By transforming subjective narratives into quantifiable network structures, this approach advances the conceptualization and measurement of travel behavior and provides affect-informed insights for active mobility management and urban governance.

Travel Behaviour and SocietyVol. 47
University of North Carolina at Chapel Hill (US), Huaqiao University (CN), Central South University (CN), McGill University (CA), University of Hong Kong (HK)
Human Mobility and Location-Based Analysis
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon