Skip to content

Using LLM technologies to analyze transport accident investigation reports

License

Notifications You must be signed in to change notification settings

1jamesthompson1/TAIC-report-summary

Repository files navigation

TAIC report analysis

Website tests codecov


A project that has built a tool and webapp to support both investigators and researchers in understanding the vast swath of previous investigations.

What

This GitHub repo is a project repository that has been built throughout my experience and efforts of applying data science solutions to transport accident investigations.

There are fundamentally two deliverables in this project.

  1. engine which is a Python program that gathers publicly available investigation reports and extracts/infers information to make datasets of safety issues, recommendations etc.
  2. viewer which is a Python Flask webapp deployed with Azure that implements a RAG search engine to search through some of the datasets generated in the engine.

There is also a collection of jupyter notebooks that show the development process of some of the more complex features in the engine and viewer.

About

This project started as a university project for my final semester. The university work was completed in July-October 2023 and finished with a basic engine and viewer app. Since then work has been completed directly with TAIC to bring the engine and viewer from POC -> Prototype -> Production. It should be released internally sometime in October 2024.

More information

If you want more information about the project and how it works/how you can use it you should checkout the wiki.

However if you are curious about the project and its history and non-technical documentation you can check out the Project documents

About

Using LLM technologies to analyze transport accident investigation reports

Topics

Resources

License

Stars

Watchers

Forks