How akashmurugesan-datascientist/geo-bias-in-llm works
This repository is a dissertation analysis project, not a deployed application. Its principal workflow compares air-quality measurements, public perceptions and labeled LLM responses, with the combined proxy dataset supporting final comparison and visualisation. The README documents the notebook’s preparation, analysis and visualisation role and names the data artifacts; no source excerpts were available, so the graph avoids claiming specific notebook operations or an exact processing order.
Main entry points: Geo_Bias_in_LLM.ipynb Analysis notebook · ground_truth_monthly_with_aqi.csv Air quality and AQI · perception_summary_monthly_complete.csv Public perception · llm_outputs_labeled.csv Labeled LLM responses
Components 6
Analysis inputs
- Air quality and AQI
ground_truth_monthly_with_aqi.csv - Public perception
perception_summary_monthly_complete.csv - Labeled LLM responses
llm_outputs_labeled.csv
Analysis workflow
- Analysis notebook
Geo_Bias_in_LLM.ipynb
Comparison results
- Comparison dataset
final_perception_proxy.csv - Dissertation findings
Geo-Bias in LLM.pdf
Connections 5
- Air quality and AQI → Analysis notebook: analysis input
README.md - Public perception → Analysis notebook: analysis input
README.md - Labeled LLM responses → Analysis notebook: analysis input
README.md - Comparison dataset → Analysis notebook: final comparison input
README.md - Analysis notebook → Dissertation findings: supports findings
README.md