Fall Symposium on
Digital Scholarship 2024

This presentation explores the development of a comprehensive digital platform dedicated to preserving and celebrating the ancient art and technology of mortise and tenon joinery—a foundational element of traditional Chinese architecture and craftsmanship spanning over 7,000 years.

Ms. Anna Qin (AVA) will demonstrate how the digital archive meticulously documents the history and evolution of mortise and tenon structures, emphasizing their seamless integration of functionality and aesthetics. The platform serves not only as a repository of 36 fundamental joint types and their applications but also as an educational tool that inspires students in creative projects focused on social innovation and service-learning. By incorporating photography, 3D models, and interactive media, the project bridges traditional craftsmanship with modern technological advancements and explores future possibilities for innovative structural designs.

Additionally, the presentation will examine the pathways this initiative creates for future design and sustainability. It will highlight how combining foundational knowledge with new technical tools empowers creators to expand the boundaries of traditional woodworking. Ms. Qin will also discuss the emerging role of Generative AI in assisting 3D design, envisioning a future where ancient techniques are revitalized through artificial intelligence to promote sustainable and aesthetically advanced architectural solutions.

“From Grain to Grid" not only preserves a rich cultural legacy but also fosters creative potential and technological progress, ensuring that the artistry and science of mortise and tenon continue to inspire and evolve in the digital era.

The use of Large Language Models (LLMs) for tasks such as idea brainstorming, code generation, and data prediction and forecasting has recently gained popularity. Thanks to their advanced natural language understanding, LLMs are able to interpret and reconstruct unclear and unrecognisable texts, making them a well-suited technology to tackle optical character recognition (OCR) tasks. 

Previous studies on OCR using LLMs have primarily focused on modern publications in Latin-based scripts. When OCRing archaic Chinese documents, which are often processed in local libraries and archives nowadays, sufficient quality has not been achieved yet, neither with conventional state-of-the-art OCR tools nor with LLM tools.

In this case study, we will discuss our investigation and application of Gemini to improve the OCR quality of typed Chinese text in archaic film publications. We will compare Gemini with other OCR tools and highlight its superior performance. We will also discuss the limitations of Gemini that we discovered, especially when the documents contain a mixture of single and multi-column text as well as complex page layouts. This case study aims to identify new opportunities to refine OCR methods and improve workflow management in the important task of preserving local cultural heritage in Hong Kong.