A Guide to Using Social Media as a Geospatial Lens for Studying Public Opinion and Behavior
2026-04-09 • Social and Information Networks
Social and Information Networks
AI summaryⓘ
The authors explain how social media and online reviews can help researchers study what people think and experience in different places. They describe a step-by-step process for collecting and analyzing this type of data, including how new language models improve understanding messy text. Four examples, like tracking COVID-19 vaccine opinions and assessing earthquake damage, show how these methods provide quick and detailed insights. Overall, the authors show that social media data can nicely complement traditional surveys and sensors for studying public opinions and local experiences.
social media datageospatial researchpublic opinionlarge language modelsdata collectioninformation extractionstatistical modelingCOVID-19 vaccine acceptanceearthquake damage assessmentplace-based experience
Authors
Lingyao Li
Abstract
Social media and online review platforms have become valuable sources for studying how people express opinions, report experiences, and respond to events across space. This work presents a practical guide to using user-generated social data for geospatial research on public opinion, human behavior, and place-based experience. It shows the promise of using these data as a form of passive, distributed, and human-centered sensing that complements traditional surveys and sensor systems. Methodologically, the chapter outlines a general workflow that includes platform-aware data collection, information extraction, geospatial anchoring, and statistical modeling. It also discusses how advances in large language models (LLMs) strengthen the ability to extract structured information from noisy and unstructured content. Four case studies illustrate this framework: COVID-19 vaccine acceptance, earthquake damage assessment, airport service quality, and accessibility in urban environment. Across these cases, social media data are shown to support timely measurement of public attitudes, rapid approximation of geographically distributed impacts, and fine-grained understanding of place-based experiences.