Tracking reports on internally displaced persons
With full scale pipeline and processing.

Features include: article retrieval, report extraction,
filtering, classification, information extraction, and visualization.

Article processing

We have built a customized web scraping infrastructure to retrieve articles and extract content into a SQL database.

Natural Language

A combination of natural language processing and keyword analysis is used to retrieve useful information from articles.

Visual analysis

A visual front-end interacts with the database to produce a map of displacement events, and highlight extracted information.

Ready to go

The tool is open source, built on Python, PostgreSQL, and Node.js, and fully Dockerized, for reliable deployment anywhere.

Access Everything

Challenge results, codebase and visualizations are all available to view.

  • GitHub Repository

    The repository that contains the open source code and notebooks for our solution.

  • Results

    A DropBox folder containing the test set classification and information extraction results.

  • Visualization Map

    Our map visualization prototype can be viewed here.

  • Documentation

    See here for how to deploy and use the code for our article processing pipeline.

Want to join the team?