D4R -- Exploring and Querying Relational Graphs Using Natural Language and Large Language Models -- the Case of Historical Documents
Project Overview
The document discusses the application of generative AI in education, highlighting the D4R platform as a transformative tool for historians that utilizes natural language processing and large language models to facilitate the exploration of historical documents. By translating user queries into graph database queries, the D4R platform allows non-technical users to engage in advanced data exploration and analysis, making it easier to navigate complex texts. This tool not only simplifies the research process for historians but also has potential applications in various educational domains by enhancing research workflows and enabling users to visualize relationships between named entities in texts. The findings suggest that incorporating generative AI in educational settings can democratize access to complex data analysis, ultimately fostering a more engaging and efficient learning experience. Through tools like the D4R platform, educators and students alike can benefit from improved data literacy and analytical skills, paving the way for innovative approaches to learning and research across disciplines.
Key Applications
D4R (for Religious Dissent and Reception of the Reformation during the Renaissance, Spain – 16th century)
Context: Historians and researchers analyzing large volumes of historical documents.
Implementation: D4R translates natural language questions into Cypher queries using a large language model and facilitates exploration through a user-friendly graphical interface.
Outcomes: Enables efficient querying of historical texts, allowing historians to uncover insights and relationships within large datasets.
Challenges: Requires an understanding of the relational graph structure; may not be accessible to all users without technical background.
Implementation Barriers
Technical Barrier
Historians may struggle with the technical aspects of querying complex datasets without prior knowledge.
Proposed Solutions: The platform is designed to be user-friendly and includes features for both non-expert and expert users.
Project Team
Michel Boeglin
Researcher
David Kahn
Researcher
Josiane Mothe
Researcher
Diego Ortiz
Researcher
David Panzoli
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Michel Boeglin, David Kahn, Josiane Mothe, Diego Ortiz, David Panzoli
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai