Semantic connection between art historical context and segmented images of paintings for virtual museum curation

Abstract

In the field of heritage data management, museums often face challenges in developing effective methods for interpreting and retrieving information for virtual curation. This is often due to a lack of integration between humanistic methodologies with computational analysis approaches. This research sought to address this issue by proposing a method of systematically designing analysis procedures based on traditional art historical practices, using a combination of topic modeling and semantic segmentation. It aims to create a semantic connection between images and text materials, enabling the creation of a semantic network that facilitated the navigation of paintings and documents for virtual art curation.