Urban Network by Haoheng

Physical Network
VS
Digital Network

Personal Project
Nov 2022 to Dec 2022
Instructor: Wenjing Li, Waishan Qiu

"Cyber Street"

0. Research Framework

Physical network or digital network,
which is more effective for the architectural style spillover effect?

Photos on social media show similarity in architectural renovation style among commercial spaces in historical areas in Shanghai even if they are far apart.

1. Similarity Network

Similarity Between Different Commercial Spaces Based on Social Media Photos and Computer Vision

Thousands of photos posted on the social media were collected. Computer vision helps make quantitative comparison between each pair of photos from different places. The aim is to calculate the similarity between each two commercial places.
For each two places, each of their photos were paired and calculated from each aspect listed above. The overall similarity is an average of all feature differences for each pair

Similarity Network Visualized on Map

Each line connects one commercial space to another.

The result of the quantitative analysis is a similarity index for each pair. The higher the index, the higher the similarity. The similarities among places form a network, which is visualized on a map.
The data visualization shows that the similarities go beyond space: even commercial spaces far apart from each other are similar in architectural styles. However, whether the physical/ digital connection between two places is high or not still needs to be tested.

2. Physical Network

2.1 Metro Commuting Network

Space connection between two places can be quantified using metro passengers' trip (OD) data and its timestamp. The more passengers going from one place to another and the less time it takes, the stronger connection are established between them.

The impact of the connections between metro stations on each shop will be quantified with an Inverse Distance Weighted (IDW) method.

MetroA→B = ∑(1/TimeA→B) = ∑(1/Timemetro + DisttoStation/ Vwalk)

15,772,842 Metro OD Data in Jun, 2016

2.2 Taxi's Ride Network

Space connection between two places can also be quantified using GPS data of taxis which take passenger from one shop to another. The less time it takes to go from one place to another, the stronger connection are established between them.

TaxiA→B = ∑(1/TimeA→B) = ∑(1/Timemetro + RBuffer/ Vwalk)

2.3 Combining Metro Commuting Network and Taxi's Ride Network

ConnectionA→B = (ln(MetroA→B + TaxiA→B)- min_val) / (max_val - min_val)*100

The values of normalized physical connection cluster around 0 and 100 because commercial spaces are either located on the same street or far apart. Data transformation using logarithm makes the values Gaussian-distributed, which facilitates comparisons in the next step

2.4 Comparing Physical Network with Similarity Network

① For these commercial spaces, the similarity network looks overlapped with the physical network.

Scatter plot shows little correlation between the physical network and the similarity network.

3. Digital Network

3.1 Weibo Users' Relationships

The distance of the shortest path from one specific node (user) to another will be calculated and used as an index for digital network connections. The shorter the path, the stronger the connection.
The Data was obtained through scraping Weibo user's fans ID and name.

3.2 Comparing Digital Network with Similarity Network

For some of the commercial spaces, the similarity network looks overlapped with the digital network.

Scatter plot shows some correlation between the digital network and the similarity network.

4. Conclusion: Digital Network > Physical Network

Methods
(1) Plot all the points in a 3D space
(2) Find out several groups through DBSCAN clustering
(3) Screen out points that represent high similarity(>70) between two places.

Results
For points whose similarity is high, more of them exist with digital connection order higher than physical connection order when physical and digital connection values are both ordered from low to high.

5. Discussion: The Image of Architecture

Simple geometries, bright colors, and reflecting materials become common in architecture, because these elements bring attractive looks in the tiny images on social media.

In the era of social media, a new requirement for architectural design emerges, which is, pursuing the spread of images on the Internet. Architects tend to create “flatten spaces”, changing 3D spaces back to 2D images to pursuit good looks in the tiny images on screens. These good-looking buildings are generally not considered as good ones since they ignore the quality and experience of space itself and even compromise on functions. However, it should be noted that the spread of images of architecture has a positive impact on the building and the city it locates in. The Guggenheim Museum in Bilbao helped widespread the image of the city. It soon became one of the most prestigious museums in Europe and attracted numerous tourists to visit Bilbao every year. Is the emphasis on images a disaster or a revolution for architecture? It is still an open question.

Physical space can no longer be considered absolute. It cannot be divorced from its digital dimension. Neither could this new system(the space of flows) be neutral.
Instead of flows replacing spaces and bits replacing atoms, cities are now the hybrid at the intersection of the two.
Carlo Ratti (2016)