Urban sentiment mapping: How AI and spatial analysis are transforming city planning

Traditional methods of understanding public sentiment in cities such as surveys and interviews often lack scale, timeliness, and objectivity. To overcome these limitations, the study introduces a two-phase computational framework combining natural language processing (NLP) and computer vision.

What if artificial intelligence (AI) could read the emotional pulse of a city? By mapping public sentiment in real-time, planners could gain unprecedented insights into how neighborhoods make people feel. With these data-driven insights, they could design cities that are not just functional but also safe, inclusive, and resilient.

In this regard, a groundbreaking study published in Frontiers in Computer Science leverages artificial intelligence (AI) to map urban sentiment using language and vision models. Titled "Urban sentiment mapping using language and vision models in spatial analysis", the study analyzes geotagged social media posts and integrating computer vision techniques on street-view imagery to offer fresh insights into how urban design shapes public emotions.

Let's dive in!

Traditional methods of understanding public sentiment in cities such as surveys and interviews often lack scale, timeliness, and objectivity. To overcome these limitations, the study introduces a two-phase computational framework combining natural language processing (NLP) and computer vision.

The first phase, called sentiment inference, applies a BERT-based deep learning model to analyze geotagged social media posts. By processing text from platforms like Instagram, the model assigns sentiment scores (either positive or negative) to each post, allowing researchers to map emotional responses to specific urban locations.

The second phase, urban context inference, utilizes advanced computer vision techniques, including PSPNet and Mask R-CNN, to analyze street-view imagery. This allows researchers to quantify key urban design features such as visual enclosure, human scale, and streetscape complexity. By overlaying sentiment data with these urban features, the study explores how different aspects of the built environment correlate with public perception.

How urban design influences public sentiment

The results reveal striking connections between urban design and sentiment. Greenery and pedestrian-friendly infrastructure consistently generate positive sentiment, while excessive openness, fenced-off areas, and high vehicle dominance tend to correlate with negative sentiment.

1. Greenery Boosts Public Happiness: The presence of trees, parks, and natural elements significantly enhances positive sentiment. This aligns with previous research on the restorative and psychological benefits of green spaces in urban settings.
2. Pedestrian Infrastructure Matters: Sidewalks and walkable spaces contribute positively to sentiment, indicating that people prefer environments designed for pedestrian movement rather than car-dominated streets.
3. Excessive Openness Is Uncomfortable: Surprisingly, too much open space such as large, empty plazas or wide roads elicits negative sentiment, suggesting that people feel more comfortable in enclosed environments that offer a sense of security and scale.
4. Fences and Walls Create a Sense of Restriction: The presence of barriers, such as fences and high walls, is associated with negative sentiment, likely due to feelings of restriction, lack of accessibility, or perceived social division.
5. Street Activity Enhances Sentiment: Areas with active streetscapes people walking, biking, or engaging in social activities correlate with positive sentiment, while inactive or deserted streets tend to evoke negativity.

Impact of societal disruptions: COVID-19 and shifting urban sentiment

The study also examined how societal disruptions, such as the COVID-19 pandemic, influenced urban sentiment. Using hotspot analysis, researchers tracked changes in emotional responses before, during, and after the pandemic.

• Pre-COVID (2019): Public sentiment largely reflected typical urban interactions, with positive sentiment clustering in areas rich in greenery and pedestrian-friendly infrastructure.
• During COVID (2020): Negative sentiment increased significantly, particularly in commercial and high-traffic areas. Lockdowns and movement restrictions contributed to public stress, and urban areas designed for social interaction saw a decline in positivity.
• Post-COVID (2021): Sentiment patterns shifted again, with positive sentiment resurging in recreational and pedestrian-friendly spaces. However, lingering cold spots in some areas indicated that pandemic-era urban challenges had long-term effects on public perception.

Designing happier cities: Key takeaways for urban planners

The findings emphasize the importance of data-driven urban design. Planners and policymakers can leverage AI-driven sentiment mapping to create environments that enhance public well-being. Key takeaways include:

• Prioritize Green Spaces: Cities should integrate trees and natural elements into urban design, not just for aesthetic reasons but for their proven psychological and emotional benefits.
• Enhance Walkability: Developing pedestrian-friendly infrastructure can significantly boost public sentiment, making cities more enjoyable and livable.
• Balance Openness and Enclosure: While open spaces are essential, they should be designed with careful consideration to avoid creating environments that feel vast and unwelcoming.
• Reduce Physical Barriers: Avoiding excessive fencing and walls in public spaces can foster a greater sense of inclusion and accessibility.
• Support Vibrant Street Life: Encouraging street-level activities such as outdoor cafes, public seating, and pedestrian zones can contribute to a more positive urban experience.

To sum up, by integrating real-time public perception into spatial analysis, cities can move beyond traditional planning methods and embrace a human-centered approach to urban development.