How Artificial Intelligence Maps Tourist Perceptions Across Shanghai’s Heritage Districts

Researchers from Tongji University used artificial intelligence to study how tourists see and feel about Shanghai’s historic neighborhoods through photos and online reviews. The study revealed that visitors favor artistic, colorful spaces but often idealize them online with cooler tones and filters, reshaping how the city is visually imagined.

Researchers from the School of Economics and Management and the College of Architecture and Urban Planning at Tongji University have developed an artificial intelligence (AI) system that can decode how tourists see and feel about historic parts of cities. Their study, “A Multidimensional AI-powered Framework for Analyzing Tourist Perception in Historic Urban Quarters,” focuses on twelve of Shanghai’s most famous heritage districts, from The Bund and West Nanjing Road to Yuyuan Road and Shanyin Road, and explores how visitors truly experience these spaces.

Historic neighborhoods, the researchers explain, are more than clusters of old buildings. They embody a city’s collective memory and identity while remaining vibrant centers of tourism, business, and daily life. Traditionally, researchers relied on surveys and fieldwork to understand public perception, but these methods are slow and limited. With the explosion of social media, people now share thousands of photos and comments about the places they visit. Tongji University’s team saw an opportunity to use AI to analyze this vast pool of data and discover what truly shapes a tourist’s experience.

Turning Social Media Into Urban Insight

To carry out the study, the researchers collected tourist photos and reviews from the popular Chinese app Dianping and compared them with real-world street views from Baidu Maps. They used a combination of AI tools, one model to analyze images, another to study colors, and a third to read emotions in text. Together, these tools formed a complete picture of how tourists visually and emotionally interact with historic Shanghai.

Their image analysis relied on advanced deep learning models like Segment Anything (SAM) and YOLOv8-Seg, fine-tuned to identify key objects in tourist photos. This allowed them to pinpoint what visitors focused on most. Another part of the framework extracted dominant color themes from the images, while a language model similar to ChatGPT studied written reviews to understand how people felt about their experiences.

What Tourists See: Buildings, Trees, and Art

The findings revealed that tourists are visually drawn to buildings above all else, accounting for nearly 20 percent of all photographed content. Trees and artworks followed closely, with walls, signs, and people also playing major roles in shaping the urban image. Each neighborhood carried a unique visual identity.

Busy commercial areas like The Bund and Wujiang Road were filled with people, vehicles, and waterfront scenes that showcased the city’s energy. Artistic quarters such as Tian’ai Road and Yuyuan Road stood out for their creative murals and public art, reinforcing their reputation as cultural hotspots. These variations, captured through AI analysis, reveal how different urban designs attract different types of attention. The study suggests that what people choose to photograph can serve as a new kind of indicator for city planners, highlighting which spaces feel lively, comfortable, or inspiring.

Color, Filters, and the “Instagram Effect”

Beyond shapes and objects, the research dove deep into color. Using a clustering algorithm and the Chinese color system, the team found that shades of orange-red and blue-violet dominate tourist photos, symbolizing warm architectural hues and cool environmental tones. Yet, a surprising discovery emerged when they compared these photos with actual street views: people tend to make their photos cooler and brighter than reality.

Social media posts were filled with blues and greens, while real-world images contained more yellows and browns. The researchers called this the “color gap”, a difference between what visitors see and what they want to see. It reflects how people use filters and editing tools to make urban scenes look cleaner, fresher, and more modern. This digital beautification creates a parallel version of the city online, one that may not perfectly match reality but strongly shapes public perception.

Reading Emotions Through Words

While images showed what tourists saw, written reviews revealed how they felt. The team trained a multi-task BERT model, a sophisticated AI language system, to analyze thousands of tourist comments. It rated satisfaction across four categories: activities, built environment, public facilities, and business options.

The results were telling. Visitors praised lively festivals and scenic walks, but often complained about overcrowding and long restroom lines. Some described certain areas as “too commercialized,” expressing frustration with rising prices and repetitive shops. On the other hand, events like the Lantern Festival in Yuyuan Garden drew enthusiastic reactions. The AI’s findings closely matched human interpretations, proving that emotional responses can be measured accurately through automated analysis.

From Algorithms to Better Cities

The Tongji University researchers believe their work could help city planners and tourism officials make better decisions. By identifying what tourists appreciate, whether it’s greenery, public art, or architectural style, urban designers can enhance spaces while preserving their historical charm. The insights also show how social media shapes the reputation of heritage sites, influencing not just how they are remembered but how they are managed.

The team acknowledges that there is room for improvement. Future versions of their AI framework could integrate more visual data and correct the “filter effect” to better match online imagery with real-life conditions. They also envision expanding the model beyond Shanghai, applying it to other world heritage cities.

Ultimately, the study blends technology with urban storytelling. It reveals how tourists perceive the city through a mix of color, culture, and emotion, and how AI can help cities listen more closely to those perceptions. By transforming social media posts into meaningful insights, the Tongji University researchers have opened a new window into the emotional life of cities, where history, technology, and human experience meet.