Assessing geographic polarisation in Britain’s digital landscape through stable dynamic embedding of spatial web data

Emerald Dilworth, Emmanouil Tranos, Daniel J. LawsonEPJ 2025 · DOI

Abstract

This paper employs Unfolded Adjacency Spectral Embedding (UASE) to investigate the temporal evolution of economic relationships between locations in Great Britain. We utilise timestamped, geolocated website hyperlinks data between archived, commercial websites in Britain, which are aggregated to create a set of directed, weighted networks of hyperlink connections between Local Authority Districts (LADs) for each year in the period 2005-2010. Thus, we are able to assess the digital evolution of longstanding economic disparities such as the North-South, Urban-Rural, and London versus the rest of the country divides in Britain. Our method is a robust and scalable statistical testing procedure for detecting changes between communities in dynamic networks where changes are expressed in terms of known covariates. We can describe network trends over time with respect to longitude and latitude covariates, relying on spatio-temporal stability properties of UASE to make comparisons of nodes in graphs over time. These trends can be formally tested with p-values as well as interpreted in terms of covariates and features of the network. We show how the methodology can be made robust to the problems of large-scale real-world data, used to detect changes over time, and identify their characteristics. This work provides the first robust evidence that commercial website hyperlink connectivity patterns between the North and South are diverging over time, highlighting an increasing digital divide.

Introduction

A central question in economic geography is whether the internet acts as a decentralizing force, redistributing economic activity across regions, or if it reinforces existing "agglomeration" effects where activity clusters in established hubs. While the "death of distance" was once predicted, empirical evidence often suggests that digital connectivity mirrors and even amplifies physical geographic divides. This paper investigates whether the expansion of the Web in Great Britain between 2005 and 2010 was associated with increased geographical polarization, specifically focusing on the North-South, Urban-Rural, and London-centric divides.

Methodology

The study models the British commercial web as a series of dynamic, directed networks. To analyze these networks, the authors employ several sophisticated statistical techniques:

Data Aggregation and Cleaning

Website-level data is aggregated into 380 Local Authority Districts (LADs). To manage the noise and extreme values inherent in web data, the authors apply log transformations and Winsorization (capping values at the 99th percentile).

Unfolded Adjacency Spectral Embedding (UASE)

UASE is used to map dynamic networks into a low-dimensional latent space. Its primary advantage is spatio-temporal stability: if a node's connectivity behavior remains constant over time, its position in the embedding space remains stable. This allows for direct comparison of regional digital behaviors across different years.

Procrustes Alignment

To interpret the abstract dimensions of the spectral embedding, a Procrustes transformation is applied. This aligns the embedding axes with real-world covariates: latitude, longitude, and the number of active enterprises. This step bridges the gap between network science and geographic analysis.

Permutation Testing

A robust statistical permutation test is introduced to determine if the distance between regional groups (e.g., North vs. South) in the aligned embedding space changed significantly between the early (2005–2007) and late (2008–2010) periods.

Data Sources

The primary data source is the JISC UK Web Domain Dataset (1996–2013), curated by the British Library. The analysis focuses on the 2005–2010 period, a time of rapid web adoption. The researchers used the Geoindex Subset to extract postcodes for geolocation and the Host Link Graph Subset for directed hyperlinks. The dataset was filtered to include only .co.uk domains with unique postcodes, serving as a proxy for local commercial businesses.

Key Insights and Results

The analysis reveals that digital topology is heavily dictated by physical geography; businesses are significantly more likely to link to others in close proximity. Contrary to the "flattening" hypothesis, the Web appears to reinforce agglomeration.

The permutation tests provided concrete evidence of diverging digital divides:

  • North-South Divide: A statistically significant divergence was found in the latitude dimension ($p = 0.0002$). While the South's digital behavior evolved and expanded rapidly, the North remained relatively static.
  • Urban-Rural Divide: Significant divergence occurred in both longitude ($p = 0.0002$) and latitude ($p = 0.0030$), indicating that rural areas were increasingly marginalized in the digital commercial landscape.
  • London vs. Rest of GB: A significant divergence was observed along the axis representing the number of active enterprises ($p = 0.0042$), underscoring London's accelerating dominance in commercial hyperlink activity.

Conclusion

This research provides robust evidence that the early expansion of the commercial web in Britain widened existing geographic inequalities. By aligning spectral embeddings with physical coordinates, the authors demonstrate that the digital landscape is not a neutral space but one that actively reflects and exacerbates regional disparities. Future work may extend this framework to multi-community analysis and incorporate industry-specific classifications to further unpack the drivers of digital polarization.

Summary updated

Abstract

Abstract This paper employs Unfolded Adjacency Spectral Embedding (UASE) to investigate the temporal evolution of economic relationships between locations in Great Britain. We utilise timestamped, geolocated website hyperlinks data between archived, commercial websites in Britain, which are aggregated to create a set of directed, weighted networks of hyperlink connections between Local Authority Districts (LADs) for each year in the period 2005-2010. Thus, we are able to assess the digital evolution of longstanding economic disparities such as the North-South, Urban-Rural, and London versus the rest of the country divides in Britain. Our method is a robust and scalable statistical testing procedure for detecting changes between communities in dynamic networks where changes are expressed in terms of known covariates. We can describe network trends over time with respect to longitude and latitude covariates, relying on spatio-temporal stability properties of UASE to make comparisons of nodes in graphs over time. These trends can be formally tested with p -values as well as interpreted in terms of covariates and features of the network. We show how the methodology can be made robust to the problems of large-scale real-world data, used to detect changes over time, and identify their characteristics. This work provides the first robust evidence that commercial website hyperlink connectivity patterns between the North and South are diverging over time, highlighting an increasing digital divide.
BibTeX
@inproceedings{dilworth2025assessing,
  author = {Emerald Dilworth and Emmanouil Tranos and Daniel J. Lawson},
  title = {{Assessing geographic polarisation in Britain’s digital landscape through stable dynamic embedding of spatial web data}},
  booktitle = {EPJ},
  year = {2025},
  doi = {10.1140/epjds/s13688-025-00537-x},
}
Figure