Diana Tay

Diana Tay (PhD in Cultural Materials Conservation, 2023) ‘Building a Conservation Material Record: A Study of Paintings by Cheong Soo Pieng and Georgette Chen’

Despite the growing visibility of prominent figures in modern Singaporean art history, there is limited material knowledge of the art practices of paintings from Nanyang artists such as Georgette Chen (1906–1993) and Cheong Soo Pieng (1917–1983). Scholarly interest in Singaporean artist materials and techniques has focused attention on the study of easel paintings through art historical and technical art history methodologies. There are art material and conservation records, but to compare these studies, the consistency and structure of the data collection make it challenging.

In response, this thesis re-assesses the development of accessible documentation methods for studying paintings by considering the type of data to be collected, the structure of the record, and whether it is possible to produce quality insights without solely relying on advanced material analysis. Employing a standard technical art history methodology, a total of 67 artworks from Cheong Soo Pieng and Georgette Chen, dated from the 1940s to the 1980s, were examined through a combination of historical and archival sources, visual examination, technical photography, and advanced material analysis. To extend the data and produce quality insights, a robust documentation record was produced where observations were recorded outlining 110 defining properties of each painting, resulting in verifiable data points to be analysed.

Differences in access to paintings and availability of archival sources affected the methodologies that could be used to gain insights. This meant that although only eight Georgette Chen paintings were studied, but nevertheless, solid correlations and consistency in canvas preparation and painting techniques were still uncovered. In a more extensive study of 59 Cheong Soo Pieng artworks, where samples were removed and analysed, the data produced improved the depth and quality of the artist’s record. In addition, the data structure of the documentation record enabled datasets to be extracted and visualised through visual graphics to uncover patterns of each artist’s art practice.

The datasets from Cheong Soo Pieng provided depth to undertake unsupervised machine learning with Principal Component Analysis (PCA) and k-means to uncover relationships from data generated from non-invasive techniques and results from the material analysis. This clustered data on Cheong Soo Pieng’s practice into four clusters, and its characteristics were examined using an interactive Microsoft Power BI dashboard.

The methodologies proposed in this study aim to build a material record of Singaporean artists that can accommodate future datasets to build onto. Using the large amounts of text and image-based data produced by this study, machine learning algorithms, including deep learning models, were explored to discover possible future uses to improve efficiencies between text and technical image diagnostics. In light of such a data rich field, the presented methodologies showcase how knowledge discovery can be accessed by employing data science methods that produce evidential, verifiable and quality data through data structure. In doing so, a better understanding of Cheong Soo Pieng and Georgette Chen’s practice is produced, contributing to the development of a robust material record and our knowledge about paintings in Singapore and potentially the wider Southeast Asian region.

Supervisors: Dr Nicole Tse, Professor Robyn Sloggett