This two-year research project aims to explore online exhibition design technologies, identify factors influencing visitor attention, and develop evaluation methods for online exhibition experiences. The first-year study (2024) focuses on examining the presentation techniques of museum and art exhibitions, conducting interviews, and analyzing multiple online exhibition experiences to identify key factors affecting visitor engagement. Through extensive comparisons and observations, we excluded subjective factors such as exhibition content preference and visual aesthetics and found that the sense of spatial presence in virtual exhibitions significantly impacts the viewing experience. Among various design elements, the degree of movement freedom in the virtual space emerged as a critical factor influencing spatial perception.

To examine this effect, we designed a within-subject experiment to test how different levels of movement freedom influence the visitor experience. A spatial perception assessment test was developed (Figure 2) to measure participants' cognitive representation of exhibition spaces, exhibit placement recognition, and memory of key exhibits. This test is applicable to both online and physical exhibitions, making it a valuable tool for comparing spatial cognition across different formats.

As of now, approximately 60% of the first-year research has been completed. Key progress includes:

Comparative Analysis of Online Exhibitions: Three exhibitions with similar curatorial scales, each with both online and physical versions, were analyzed (Figure 1). The comparative study has been submitted to an international journal.

Movement Freedom in Virtual Spaces: Three different online exhibition platforms, each utilizing different construction techniques and varying levels of movement freedom, were selected for experimental analysis (Figure 3). The impact of navigation constraints on visitor experience and spatial perception was assessed.

AI-Based Attention Prediction: Various machine learning and deep learning models, including Attention Insight and EyeQuant, were tested (Figure 4). These tools were applied to multiple online exhibitions, such as the VR exhibition of the National Palace Museum and the Metropolitan Museum of Art, to predict visitor attention distribution in virtual exhibition spaces.

These findings contribute to a data-driven understanding of online exhibition experiences and provide scientific measurement tools for exhibition design. The study will continue with further experiments and validation to optimize interactive engagement and visual appeal in online exhibitions.