All Foreign Gifts Around Us (github.com)

A new project titled "All Foreign Gifts Around Us" has been launched, utilizing Large Language Models (LLMs) like Claude 3 Sonnet and Claude 3 Haiku to extract structured data from Federal Register publications detailing gifts received by U.S. officials from foreign sources. This initiative aims to convert unstructured text into JSON format, making the data more accessible for analysis and visualization. The significance of this project lies in its potential to enhance transparency regarding foreign influence in U.S. governance by systematically documenting these gifts, which are disclosed at a minimum value of $480. The extraction process includes various stages such as PDF conversion, structured data extraction, and database creation. It ensures data quality through de-duplication and anonymization, and outputs data in multiple formats, including JSON, CSV, and SQLite databases. This comprehensive approach to documenting foreign gifts not only aids in tracing potential conflicts of interest but also offers valuable insights through built-in analytics and visualizations. Furthermore, the project's open-source nature allows for community contributions and future upgrades, such as a web dashboard and an API for easier programmatic access to the data.