Related Experiment Video
Updated: Jul 17, 2025

Evidence-based Knowledge Synthesis and Hypothesis Validation: Navigating Biomedical Knowledge Bases via Explainable AI and Agentic Systems
Published on: June 13, 2025
Investigating antiquities trafficking with generative pre-trained transformer (GPT)-3 enabled knowledge graphs: A
Shawn Graham1, Donna Yates2, Ahmed El-Roby3
1Department of History, Carleton University, Ottawa, Ontario, Canada.
We used a large language model (GPT-3) to semi-automate knowledge graph creation for the antiquities trade, achieving comparable results to manual methods with significant time savings. This approach enhances the study of archaeological data.
Area of Science:
- Digital Humanities
- Archaeology
- Artificial Intelligence
Background:
- The antiquities trade generates vast amounts of unstructured data across diverse sources like articles, auction catalogs, and archives.
- Systematic examination of these sources is crucial for understanding the trade but is labor-intensive.
- Knowledge graphs offer a structured way to represent complex relationships within this data.
Purpose of the Study:
- To explore the efficacy of using a large language model (GPT-3) for semi-automating knowledge graph construction from scholarly texts on the antiquities trade.
- To compare the quality and efficiency of AI-generated knowledge graphs against manually created ones.
- To assess the potential of this method for discovering new insights into the antiquities trade.
Main Methods:
- A prompt-guided GPT-3 was employed to extract subject-predicate-object relationships from articles, forming a knowledge graph.
- The AI-generated knowledge graph was compared to a manually annotated version from the same source material.
- Knowledge graphs were projected into a neural network embedding model (Ampligraph) to identify probable connections and relationships.
Main Results:
- The semi-automatic knowledge graph generation using GPT-3 yielded results comparable to the manually created knowledge graph.
- This AI-driven approach demonstrated substantial time savings compared to manual annotation.
- The method allows for a potential expansion of the volume of materials that can be analyzed.
Conclusions:
- Semi-automating knowledge graph creation with large language models is a viable and efficient method for analyzing the antiquities trade.
- This computational approach has significant implications for processing archaeological knowledge found in grey literature and other scholarly formats.
- The methodology offers a scalable solution for uncovering new research avenues in archaeology and related fields.
Related Concept Videos
Spanning Openings in Brick Walls
Lintels are primary supports used to span openings and can be crafted from materials such as reinforced concrete, steel-reinforced brick masonry, or simple steel angles. These are straightforward to install and are typically concealed...
Archival Research
Ogive Graph
Extraction: Advanced Methods
Quarrying of Stone
One common method involves using a diamond belt saw to cut large blocks from the quarry face. These blocks can be about 50 feet long and 12 feet high. After the initial vertical cut, drilling is performed at the base of the...
Theorems of Pappus and Guldinus: Problem Solving

