Colonia Dignidad Files Project
2024–2025Application of methodologies to analyze the intelligence archive of the former Colonia Dignidad: we transformed 46,308 intelligence files from the dictatorship into a massive knowledge graph with almost 64,000 entities: 35,000 of them People, and 124,000 relationships.
Project
Developed in the context of a collaboration between the IMFD and the National Human Rights Institute (INDH), this project consists of two stages. First, the automatic creation of a knowledge base stored as a graph from a body of documents corresponding to intelligence files found in Colonia Dignidad, previously transcribed by the INDH, and then, a graphical interface with the aim of facilitating the exploration, consultation, and interpretation of this data. The system processes these documents, automatically extracts entities of interest (e.g., People, places, etc.) using automatic entity recognition (NER) tools, and links them together. The result is stored in our MillenniumDB graph management system. These links are then presented in a visualization, allowing entities and connections to be explored intuitively.
The application offers a query section that allows you to write and execute queries directly, displaying the results in an interactive table. Another feature is the paths section, which allows you to find relationships between a set of nodes based on certain user-defined parameters. The frontend was built with React and Vite, using Material UI (MUI) for the design, along with visualization tools such as React Force Graph, AG Grid, and Monaco Editor.