Speaker
Description
Background: Historic districts can offer positive emotional experiences to the public (Scopelliti et al., 2019; Reece et al., 2022). Visual perception is one of the most direct ways people experience historic districts, so the visual landscape characteristics of historic districts may influence public emotions. However, emotional experiences and visual landscape characteristics have often been studied as two separate themes, and the impact of visual landscapes in historic districts on emotional experiences has yet to be revealed. Artificial Intelligence methods, particularly multimodal large models can provide an approach for analyzing the relationship between visual landscapes and public emotions. Therefore, this study aims to construct a multimodal deep learning framework to explore the relationship between visual landscapes and public emotions in historic districts.
Methods: Text reviews from social media help understand people’s emotional experiences with the built environment (Ghahramani et al., 2021; Park et al., 2020), and images voluntarily shared on social media serve as a rich resource for understanding people’s visual preferences for environments (Chen et al., 2020). Although some studies have attempted to qualitatively explore the relationship between visual perception and emotional experience (Yang & Zhang, 2024), no quantitative analysis framework for this relationship has been established. In this study, we use Nanluoguxiang, a typical historic district in Beijing, as a case study. Drawing on 5,251 text reviews and 23,553 associated images from Dianping between 2022 and 2024, this study develops a multimodal deep learning framework that integrates GPT-4, U-net, multi-layer attention mechanisms, and interpretability analysis modules (SHAP and Grad-CAM) to examine the impact of 8 environmental scenes and 14 environmental elements in historic districts on public emotional experiences.
Results: The results show that: (1) Public emotions in Nanluoguxiang are predominantly positive, with a complex emotional state during the pandemic; (2) Architectural and natural landscape scenes are more strongly associated with positive emotions, while street scenes are linked to negative emotions; (3) For positive emotions, the most significant contributions come from architectural elements such as tiles and plants, as well as traditional food and handicrafts within commercial scenes; for negative emotions, the greatest influence is exerted by commercial signage in street scenes.
Conclusion: This study reveals that certain types and specific elements within historic districts impact public emotions. At the same time, the multimodal deep learning framework effectively predicts the impact of visual landscapes on public emotions in historic districts.
References
Chen, M., Arribas-Bel, D. and Singleton, A., 2020. Quantifying the characteristics of the local urban environment through geotagged Flickr photographs and image recognition. ISPRS International Journal of Geo-Information, 9(4), p.264.
Ghahramani, M., Galle, N.J., Ratti, C. and Pilla, F., 2021. Tales of a city: Sentiment analysis of urban green space in Dublin. Cities, 119, p.103395.
Park, S.B., Kim, J., Lee, Y.K. and Ok, C.M., 2020. Visualizing theme park visitors’ emotions using social media analytics and geospatial analytics. Tourism Management, 80, p.104127.
Reece, R., Bornioli, A., Bray, I., et al., 2022. Exposure to green, blue and historic environments and mental well-being: a comparison between virtual reality head-mounted display and flat screen exposure. International Journal of Environmental Research and Public Health, 15, p.9457. DOI: 10.3390/ijerph19159457.
Scopelliti, M., Carrus, G. and Bonaiuto, M., 2019. Is it really nature that restores people? A comparison with historical sites with high restorative potential. Frontiers in Psychology, 9, p.2742. DOI: 10.3389/fpsyg.2018.02742.
Yang, C. and Zhang, Y., 2024. Public emotions and visual perception of the East Coast Park in Singapore: A deep learning method using social media data. Urban Forestry & Urban Greening, 94, p.128285.
| Keywords | Artificial Intelligence; Deep learning; Emotions; Historic districts |
|---|---|
| Best Congress Paper Award | Yes |
