2026-09-09

7 arXiv categories· 96 journals· 2576 candidates — 10 worth your time· 192 without an open abstract

Headline

Narrative structure in story maps, evaluated for user engagement and trust

Today, together

tag shift
no tag ran above its 30-day average
on the same shoulders
5 of the day's papers cite Travel demand and the 3Ds: Density, diversity, and design (Robert Cervero, Kara M. Kockelman 1997) counted over the 107 of this day's 202 named papers whose reference lists we have — standing on the same older work is not the same as being about the same thing
first appearance
Decision Systems (United States); Hanbat National University; Instituto Geográfico Nacional; and 3 more none of them on a paper we published in the 89 days before this issue

Two counts, not one counted twice: every row here but “on the same shoulders” counts the papers published above, and that one counts every paper this day named — those and the ones under Also published today, whose abstracts we could not read.

This study establishes a conceptual framework for defining and measuring user engagement with open-data-based geovisualizations. Another empirical map study evaluated how narrative structure affects user engagement, performance, and trust by testing 135 participants recruited via Prolific. Finally, a new publication presents the updated printed edition of Part VI of the National Atlas of Spain, focusing on social services and facilities across three thematic chapters.

codedatapublished: Abstracts of the ICA

Maria Caudevilla-Lambán, Timothy Prestby, Gareth Baldrica-Franklin, et al.
Abstracts of the ICA

Universidad de Zaragoza

Media Influence and Health · Identity, Memory, and Therapy · Creativity in Education and Neuroscience · survey · Universidad de Zaragoza · James Madison University

The Influence of Narrative Structure on Story Map Engagement, Performance, & Trust: A Preliminary Study

This empirical map study evaluated how narrative structure affects user engagement, performance, and trust by testing 135 participants recruited via Prolific. Using a custom-built 'Story' module in the MapStudy framework, the researchers compared two variations of the 'martini glass' narrative structure: placing an interactive map at the beginning (reverse martini glass) versus at the end of a linear, scroll-based story. The experiment utilized two environmental hazard stories—one on Hurricane Harvey in Texas and another on the Tubbs Fire in California—and tracked element-level viewing times and map interactions.

Why it matters — It provides a framework and empirical baseline for testing how the sequencing of author-driven scrollytelling and reader-driven map exploration influences how audiences comprehend and trust spatial narratives.

Caveat: The study is presented as a preliminary investigation with a relatively small sample size of 135 participants split across three geographic groups.

Abstracts of the ICA · doi · code

codedatapublished: Transportation Research Part B Methodological

Chengyuan Zhang, Zhengbing He, Cathy Wu, et al.
Transportation Research Part B Methodological · was arXiv

McGill University

Autonomous Vehicle Technology and Safety · Traffic control and management · Traffic Prediction and Management Techniques · deep learning · McGill University · University of Nottingham Ningbo China

When context is not enough: Modeling unexplained variability in car-following behavior

This study introduces a stochastic framework that combines deep neural networks with a nonstationary Gaussian process to model unexplained variability in car-following behavior. Tested on the naturalistic HighD German highway dataset, the model uses a scenario-adaptive Gibbs kernel to quantify uncertainty in acceleration, speed, and spacing based on driving context.

Why it matters — It provides a method to measure the temporal persistence and magnitude of residual driving behaviors, enabling more realistic, uncertainty-aware microscopic traffic simulations and safety evaluations.

Caveat: The framework's performance and parameters were evaluated using a single dataset of highway driving, which may not represent urban or stop-and-go traffic conditions.

Transportation Research Part B Methodological · doi · code

datapublished: Abstracts of the ICA

Irene Sahagún Luis
Abstracts of the ICA

Instituto Geográfico Nacional

spatial analysis · census data · Instituto Geográfico Nacional

New printed edition of the National Atlas of Spain. Part VI. Social services and facilities

This publication presents the updated printed edition of Part VI of the National Atlas of Spain, focusing on social services and facilities across three thematic chapters: education, science, culture, and sport; health and social policies; and defence, home affairs, and justice. The atlas incorporates a new double-page spread comparing Spain's national indicators against the broader European framework and integrates gender-disaggregated official data across its thematic maps.

Why it matters — It provides a standardized, cartographic baseline of Spain's social infrastructure and welfare distribution, allowing planners to analyze regional disparities in public service access with integrated gender equality indicators.

Abstracts of the ICA · doi

datapublished: Abstracts of the ICA

Andrea Miletić, Ana Kuveždić Divjak
Abstracts of the ICA

University of Zagreb

Geographic Information Systems Studies · Data Visualization and Analytics · Spatial Cognition and Navigation · University of Zagreb

Towards a Framework for Measuring User Engagement in Geovisualizations Based on Open Data

This study establishes a conceptual framework for defining and measuring user engagement with open-data-based geovisualizations. It synthesizes literature from Human-Computer Interaction and spatial data science to address inconsistencies in how engagement is distinguished from related concepts like user experience, interest, and sustained attention.

Why it matters — It provides a structured foundation for evaluating how citizens interact with spatial open data, moving the metric of open data success from the sheer volume of published datasets to the quality of user comprehension and interaction.

Caveat: The framework is conceptual and the abstract does not indicate that it has been empirically tested or validated on specific user groups or geovisualization platforms.

Abstracts of the ICA · doi

published: Journal of Housing and the Built Environment

Hojun Lee, Hoon Han, Chris Pettit, et al.
Journal of Housing and the Built Environment

UNSW Sydney

Housing Market and Economics · Spatial and Panel Data Analysis · Land Use and Ecosystem Services · spatial econometrics · random forest · housing transactions

Geographical variations in AVM accuracy: comparing statistical and machine learning models in Greater Sydney

The study evaluated the spatial accuracy of five automated valuation models—ordinary least squares, geographically weighted regression, random forest, extreme gradient boosting, and multilayer perceptron—across housing submarkets in Greater Sydney. Random forest achieved the highest overall predictive accuracy, while geographically weighted regression performed competitively only in stable submarkets characterized by homogeneous housing and high transaction volumes. Correlation analysis showed that model accuracy systematically improves in areas with lower average housing prices, lower price dispersion, and higher sales volumes.

Why it matters — It demonstrates that automated valuation model performance is highly sensitive to local market conditions, showing that practitioners cannot rely on a single universal model and should instead tailor model selection to specific submarket characteristics.

Caveat: The findings are based on a single metropolitan area, and the specific performance thresholds of the models may differ in other urban housing markets.

Journal of Housing and the Built Environment · doi

preprint

Yi Wang, Jing Li, Jinliang Deng, et al.
arXiv cs.LG

United States · China

Inferring Urban Mobility Interactions from Aggregated Dynamics

An uncertainty-aware, physics-informed framework reconstructs temporal origin-destination (OD) matrices using only aggregated, area-level population counts. Tested across twelve mobility datasets from cities in the United States and China, the model achieves prediction accuracy comparable to methods requiring historical OD matrices as input, while correcting the systematic underestimation of sparse, high-volume corridors.

Why it matters — It demonstrates that directional flow patterns can be accurately inferred without continuous, individual-level tracking, offering a privacy-preserving and less resource-intensive method for real-time urban mobility monitoring.

arXiv · pdf

preprint

Namwoo Kim, Jeeyun Chang, Kanghoon Lee, et al.
arXiv cs.LG

New York City · Chicago

Synergistic Fusion of Topological Structure and Temporal Semantics of Mobility for Urban Region Embedding

The researchers developed Mobility Stream-Structure Synergy (MoSS), a region embedding model that represents urban areas by combining hourly inflow/outflow profiles with topological structure views derived from zigzag persistence diagrams. Evaluated in New York City and Chicago, the model captures how regional connectivity changes over time and extracts higher-order signals from the co-occurrence of these temporal and structural views.

Why it matters — It demonstrates that modeling the temporal dynamics and topological structure of human mobility alone can outperform existing region representation methods that rely on auxiliary data modalities, simplifying the data requirements for urban sensing tasks.

Caveat: The model's performance was validated on downstream tasks in only two major US cities, New York City and Chicago, which may not represent mobility dynamics in smaller or differently structured urban areas.

arXiv · pdf

preprint

Mohamad Alkadamani, Halim Yanikomeroglu
arXiv eess.SY

TD-STGT: A Spatio-Temporal Graph Transformer for Mobile Traffic Demand Forecasting

The Traffic Demand Spatio-Temporal Graph Transformer (TD-STGT) is a new graph neural forecasting framework designed to predict grid-level changes in wireless mobile traffic demand. The model utilizes a population-scaled demand proxy constructed from daytime population data and crowdsourced mobile measurements. Evaluated across five Canadian metropolitan regions, TD-STGT outperformed the strongest baseline model, achieving a ΔR² of 0.462 and reducing ΔRMSE by 5.7%.

Why it matters — It provides network operators with a highly accurate, fine-grained forecasting tool to proactively identify localized demand surges, allowing them to prioritize infrastructure upgrades like 5G/6G site densification and spectrum activation before congestion occurs.

Caveat: The model relies on a population-scaled proxy for demand rather than direct, proprietary carrier traffic measurements.

arXiv · pdf

preprint

Zijian Shen, Bin Zhou, Jiguang Wang, et al.
arXiv cs.AI

large language model · Hong Kong

LEBGen: An LLM-Enhanced Bayesian Network Framework for Few-Shot Travel Survey Data Generation

The researchers developed LEBGen, a framework that combines large language models with Bayesian networks to generate synthetic travel survey records from small samples. Tested on a 2% sample of the 2022 Hong Kong Travel Characteristics Survey, the model reduced the mean marginal Jensen-Shannon divergence from 0.0671 to 0.0091 and decreased the mean absolute Cramer's V error by 14.3% compared to the strongest baseline.

Why it matters — It enables planners to generate highly accurate synthetic travel datasets even when only a tiny fraction of representative survey data is available, overcoming the data-scarcity constraints of traditional probabilistic models by injecting behavioral knowledge from language models.

Caveat: The framework's performance was validated using a single dataset from Hong Kong, meaning its effectiveness across different cultural or regional travel behaviors remains to be tested.

arXiv · pdf

preprint

Juncen Long, Xiaofeng Jin, Gianluca Bardaro, et al.
arXiv cs.CV

MamMA: A Mamba-Based Pedestrian Trajectory Prediction Algorithm Considering Occupancy Map and Pedestrian Awareness States

The MamMA algorithm predicts pedestrian trajectories by combining local occupancy maps from LiDAR with egocentric vision data that captures pedestrian awareness states. It processes the occupancy maps in patches to extract obstacle features and integrates these with behavioral cues using a Mamba-based sequence model. The model was evaluated on five datasets: STCrowd, SiT, JRDB, ETH, and UCY, outperforming existing state-of-the-art methods in average and final displacement errors.

Why it matters — It bridges the gap between theoretical trajectory models and practical robot deployments by utilizing the actual sensor modalities available on mobile robots—specifically LiDAR occupancy maps and egocentric cameras—rather than relying on unrealistic top-down aerial views.

arXiv · pdf

Still cited

Sherwin Rosen (1974), Hedonic Prices and Implicit Markets: Product Differentiation in Pure Competition 1 of today's items cite it · 21 of 3240 in the archive stand on it

Also published today

These appeared today in journals we track. Their abstracts are not openly available, so we cannot summarise them.