Nuestra MemorIA

  • Años: 2024 – Ongoing

Nuestra MemorIA is a transdisciplinary research initiative that seeks to develop an innovative framework for analyzing the archives of the Chilean civic-military dictatorship (1973–1990) through the use of new technologies.

Proyecto

NuestramemorIA emerged as a response to the challenge faced by research teams attempting to process thousands of documents, photographs, and audio recordings from the period—materials that are currently scattered, fragmented, or in an advanced state of deterioration.

The project is funded by ANID through the Exploratorio 2025 grant (13250134) and Fondecyt (1251890).

Unlike commercial Artificial Intelligence (AI) tools, NuestramemorIA transforms static, physical collections into dynamic digital datasets that can be accessed remotely and efficiently. This facilitates the identification of patterns and connections that would be impossible to detect manually.

NuestramemorIA employs a collaborative methodology: the process is divided into five critical stages that combine the power of computational processing with human expert judgment.

I.—From image to text (OCR): We clean and digitize damaged files. This involves creating the DHiSS dataset, which currently contains over 160,000 verified word images, to train models capable of understanding the language and formatting characteristic of the era.

II.—From text to entities (NER): We automatically identify key names, dates, and locations within the vast collection of documents.

III.—Graph construction: We connect these entities. For instance, if a name appears in both a document and a judicial file, the system establishes a link between them.

IV.—Verifiable query system: The AI ​​acts as a “forensic partner”— when a user submits a query, the system provides the requested information along with a “pointer” or link to the original document, thereby eliminating AI “hallucinations.”

V.—Human validation (human-in-the-loop): Experts in history and law oversee and validate every step of the technical process, ensuring that the technology remains faithful to the historical context.

Equipo de desarrollo

  • Jocelyn Dunstan:
    Associate Professor in the Department of Computer Science and the Institute for Mathematical and Computational Engineering at the Pontifical Catholic University of Chile. She is also a researcher at the IMFD, AC3E, and iHealth centers. Her work focuses on two main areas: the use of artificial intelligence to analyze historical archives related to human rights and transitional justice, and the development of clinical natural language processing in Spanish. In both fields, she leads transdisciplinary projects that incorporate human involvement and prioritize privacy, operating at the intersection of academic research and public policy.
  • Hugo Rojas:
    Holds a PhD in Sociology (University of Oxford) and a PhD in Law (University of Salamanca). His work lies at the intersection of human rights, transitional justice, archives, and state violence in Latin America. He is an academic at the Catholic University of Uruguay, Alberto Hurtado University, and the VioDemos Millennium Institute. He also serves as an associate researcher at the Ibero-American Observatory on Criminal Justice at the University of Salamanca and as editor of the *Revista de Derecho* (Law Journal) at the Catholic University of Uruguay.
  • Domingo Mery:
    Holds a PhD in Engineering Sciences (Technical University of Berlin). He is an academic in the Department of Computer Science at the Pontifical Catholic University of Chile and a researcher at the Millennium Institute for Foundational Research on Data (IMFD). He is an expert in computer vision, image restoration, and facial recognition in low-quality images—skills that are essential for analyzing documents and photographs from historical archives.
  • Pablo Seguel:
    Holds a PhD and a Master’s degree in History from the University of Santiago, Chile, and a degree in Sociology from the University of Chile. He specializes in contemporary history, memory, heritage, archives, and human rights. He served as Coordinator of the Memory and Human Rights Unit at the National Monuments Council (2015–2022). He has extensive experience in teaching and research at both national and international universities and institutions. He is currently a member of the *NuestramemorIA* project, serving as a postdoctoral researcher at Alberto Hurtado University.
  • Felipe Mallea:
    Sociologist (Alberto Hurtado University) and holder of a Master’s degree in Social Research Methods (Diego Portales University). A researcher specializing in human rights, social memory, and the representation of the recent past in Chile, with a focus on museums, sites, and archives from the dictatorship era.
  • Juan Reutter:
    PhD in Computer Science (University of Edinburgh). Director of the Millennium Institute for Foundational Research on Data and Associate Professor in the Department of Computer Science at the Pontifical Catholic University of Chile. An expert in data management, the web, and graph databases.
  • Camilo Gutiérrez:
    Electrical Civil Engineer from the Pontifical Catholic University of Chile. He specializes in machine learning and signal processing, creating artificial intelligence architectures for the analysis of complex, unstructured data. He also develops advanced retrieval-augmented generation (RAG) systems and tools based on large language models (LLMs) for structuring large-scale databases.
  • Paula Mónica Silva:
    Journalist with a degree in Social Communication from the University of Santiago, Chile. She specializes in organizational, scientific, and data science communication. Her work focuses on the management, strategic planning, and dissemination of institutional content, bridging innovation, science, education, and the social appropriation of knowledge.