This study aims to explore the potential of the metaverse to determine if it can be an effective medium for realizing an end-user-driven heritage interpretation platform. To this end, we conducted an in-depth user-study using a combination of qualitative experimental methods (cultural probes and a focus group interview) to obtain data on the thought processes, experiences, and responses of end-users within the metaverse called the Peaceful Hill, a heritage metaverse created by the National Museum of Korea and hosted on Zepeto, a metaverse platform. The twelve participants in their 20s were recruited for a five-day experiment. Through an inductive thematic study, four themes related to the end-user experience in the process of virtual heritage interpretation were identified, as follows: structural components as virtual wider settings, interactive components as life-logging experience, mediating components as an actor, and heritage interpretation components. Then, based on the visual network analysis, we further explored the multiple connections and reciprocal influences of the above themes to uncover differences and similarities in user-perceived experiences in the metaverse compared to existing virtual environments. In addition, we offered insights and design considerations for establishing a participatory heritage interpretation experience in the metaverse.
We propose a framework based on the 5W1H model-based metadata for remote diagnosis in virtual reality (VR). For this purpose, we suggest unique metadata composed of Point of Interest (POI)-extended anchor (xAnchor)-content for a context-aware service in virtual and augmented reality. We define the attributes of the metadata based on the 5W1H context for information retrieval according to the context in a remote diagnosis. Second, we propose the ontology-based linker metadata that express the relations between AR scenes and that retrieve external information. Moreover, we suggest heritage building information metadata for information retrieval according to context. For evaluation, we created a geo-tagged content tool and a remote diagnosis VR application. We conducted focus-group interviews and heuristic evaluations for remote diagnosis in VR to verify the methodology of this study. As a result, we found that experts were most satisfied with the functions that provide the contextualized information. This study contributes to the geospatial metadata for a context-aware service in VR/AR as well as the remote diagnosis framework to overcome the time-consuming problem of the existing remote diagnosis.
3 October 2025 in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Hallucination problem is the main cause of the weak reliability of Large Language Models (LLMs) for their use in cultural institutions, such as museums and galleries. One proposed solution to the hallucination problem is to ground the LLM in the real data found on the Web. However, since the cultural heritage domain requires factual accuracy, cultural institutions cannot fully rely on the data obtained from the Web. To make the data suitable for the heritage domain use case, additional source filtering and verification must be applied. In this paper, we propose a potential source verification pipeline for verifying web sources, as well as a question-generating agent designed to guide heritage experts in collecting the right sources for their needs. Upon evaluation, the proposed system successfully filters the web-scraped sources given a search keyword, achieving moderate results in both classification tasks. In addition, our contributions include the curation of a custom dataset for training both models and estimation of an optimal training & dataset configuration for the proposed ’curatorial question generation’ task.
1 October 2025 in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Stone inscriptions pose distinct challenges for optical character recognition due to erosion, disruption, and fragmentation. We present InscriberNet, a deep learning network designed for the recognition of Chinese characters from degraded stone surfaces. It incorporates a CNN-based denoising module, a ResNet-based feature extractor, a Multi-Scale Attention Mechanism, and a Swin Transformer to capture both local and global data. InscriberNet, assessed using the Hwaeom Stone Sutra dataset, attains an accuracy of 86.5%, surpassing conventional models such as CRNN and CNN-based denoising OCR. The findings underscore its resilience, efficacy, and relevance for the digitization and preservation of culturally important historical documents.