Enhancing Geospatial Analysis through Geonarratives: An Innovative Software Approach for Health-Related Phenomena — 24a — Kaden Bretzman

Background: Geonarratives integrate qualitative narratives with geospatial data, providing spatially relevant commentary that enhances understanding of social contexts, particularly in health-related phenomena. Traditional geospatial data processing often overlooks these nuanced contexts. This project addresses this gap by developing software to convert narrative text and
GPS data into spatial formats, facilitating a more comprehensive analysis of health-related and social phenomena.

Goal: The primary objective is to advance geospatial data processing by integrating nuanced social contexts into health-related analyses through the development of efficient software for converting narrative text and GPS data into spatial formats

Methods: The process involves recording synchronized audio and GPS data, using NLP techniques in Python for transcription and linking narratives to specific coordinates. The resulting CSV file is processed using Pandas to handle geospatial and textual data. A geographical map displays timestamps and narratives. Textual data undergoes preprocessing, including lowercasing, tokenization, and filtering. Sentences are transformed into numerical vectors using TF-IDF Vectorization, and a cosine similarity matrix is computed to construct a graph. The PageRank algorithm ranks sentences by importance for summarization, and keywords are identified using TF-IDF.

Results: The software generates interactive maps and word clouds from uploaded narratives and GPS coordinates. NLP techniques derive themes from specific locations. The maps, downloadable as HTML files, display narrative commentary at specific GPS coordinates, enabling spatially distributed commentary analysis. Word clouds highlighting common themes
within specific timestamps or locations are also produced.

Conclusion: This research demonstrates the software’s potential to enhance geospatial research methodologies. By integrating narrative text with geospatial data, the software facilitates a comprehensive understanding of complex social and environmental processes, enriching the analysis of health-related phenomena and offering new insights into spatial-temporal patterns of human experiences.

University of South Dakota
Lisa McFadden