Geo-spatial Mapping of Sentiment Analysis with Transformer-Based Models

Author

Dugan J. Turnbow

Date of Award

3-2025

Document Type

Thesis

Degree Name

Master of Science in Operations Research

Department

Department of Operational Sciences

First Advisor

Phillip M. LaCasse, PhD

Abstract

The public sentiment of events of interest, and their impacts, is vital for decision makers to allocate resources. This research develops a robust algorithm for aggregating sentiment analysis from social media and published articles, while contextualizing results through spatial and temporal mapping. The methodology employs two transformer-based language models for sentiment analysis and named entity recognition (NER). Sentiment scores are generated and augmented using explicit location data, such as latitude and longitude, and implicit location data derived through NER or location features. Results are mapped using a geo-tagged location dictionary, enabling visualization of sentiment trends at state and county levels over time.

AFIT Designator

AFIT-ENS-MS-25-M-248

Comments

An embargo was observed for posting this work.

Distribution Statement A: Distribution Unlimited. Approved for public release. PA case number: 88ABW-2025-0309

Recommended Citation

Turnbow, Dugan J., "Geo-spatial Mapping of Sentiment Analysis with Transformer-Based Models" (2025). Theses and Dissertations. 8221.
https://scholar.afit.edu/etd/8221