The starting point of the study is an input–output model, a framework borrowed from economics in which a system is described as a set of components that feed into and draw from one another. Using this approach, the team identified fourteen key components of smart cities, ranging from data assets, information and communication technology infrastructure, and smart mobility to human capital, governance arrangements, and the natural environment. The choice of an input–output structure is deliberate: rather than listing smart city features in isolation, the model forces every component to be positioned in relation to the others, capturing the idea that a city becomes smart not because of any single technology but because of the way technologies, people, and institutions interact.

From those fourteen components, the researchers extended the model using a social network framework. In network terms, each component becomes a node, and every possible relationship between two components becomes a potential edge. With fourteen nodes, the mathematics of complete graphs yields ninety-one possible pairwise connections, and the authors mapped all of them as potential interconnections within the smart city system. This is where the study departs from most earlier conceptual work. Traditional smart city frameworks, such as the widely cited integrative framework of technology, people, and institutions, tend to group components into categories and describe their relationships in qualitative terms. A network representation goes further: it makes the relationships themselves objects of measurement, so that researchers can ask not only which components matter but which connections matter, and how both change over time.

To move from a theoretical network to an evidence-based one, the team turned to bibliometrics, the quantitative analysis of scientific publications. They searched the literature from 1998 to 2023, a twenty-five-year window that covers essentially the entire history of the modern smart city debate, and retrieved 214,211 articles. For each of the ninety-one potential edges in the network, they counted how often the corresponding pair of components appeared together in the smart city literature, using carefully structured search queries that combined the terms smart city or smart cities with the keywords of each component pair. The volume of co-occurrence served as a proxy for the significance of each connection: an edge that appears in tens of thousands of publications is clearly more central to how scholars and practitioners think about smart cities than an edge that appears in only a handful.

The results of this quantitative mapping are striking. The analysis highlights the central role of data assets, which emerge as the most significant node in the network, connected to nearly everything else. This finding reflects a broader transformation in how cities are imagined and managed. In the early years of the smart city debate, around the late 1990s and 2000s, discussions centered on information and communication technology infrastructure and e-government, echoing the maturity models of digital government research. As sensor networks, mobile devices, and urban data platforms proliferated, data itself became the connective tissue of the smart city, the resource that links mobility systems to energy grids, public services to citizen engagement, and environmental monitoring to policy decisions. The bibliometric evidence shows this shift happening in the publication record itself, with data-related connections growing in prominence as the decades progress.

The study does not stop at counting. The authors complemented the bibliometric analysis with a qualitative review of the twenty-five most relevant articles in the corpus, focusing on the interplay among three main components of the network. This mixed-methods design matters because raw co-occurrence counts can be misleading: two terms may appear together frequently simply because both are fashionable, not because the underlying relationship is well understood. By reading the most influential papers closely, the researchers were able to interpret what the strongest connections in the network actually mean in practice, how they are theorized, and where the literature reveals genuine integration versus superficial association. The qualitative layer also helps explain the dynamics behind the numbers, showing how the meaning of individual components has evolved as the smart city concept has matured from a technology-driven vision into a more multidimensional model of urban governance.

One of the study’s central conclusions is that smart city development is dynamic and multidimensional in a way that static frameworks fail to capture. The network is not fixed: nodes gain and lose importance, and the edges between them strengthen or weaken as technologies mature, policies change, and societal priorities shift. For example, the relationship between data assets and human capital or intellectual capital has become increasingly visible in recent years, reflecting debates about the skills, privacy protections, and institutional capacities cities need to turn raw data into public value. Similarly, connections involving governance and sustainability have grown as critics have pushed back against purely technological visions, asking whether a city can be smart without being sustainable, inclusive, or democratic. The network model makes these evolutions measurable, allowing researchers to track them across successive time windows rather than relying on impressionistic assessments.

For practitioners, the implications are concrete. City officials and urban planners often face difficult decisions about where to invest limited resources: in sensors and platforms, in skills and organizational change, or in partnerships with private providers. A network-based view of the smart city offers a way to ground those decisions in evidence about which components and connections the accumulated research literature considers most consequential. It also provides a diagnostic tool: a city that has invested heavily in infrastructure but neglected the human capital and governance nodes may find that its data flows never translate into better services, because the edges connecting technology to people and institutions are weak. The authors suggest that the model can help researchers and practitioners alike pay attention to changes in the smart city concept over time and to the connections between the various components involved in the development process, rather than treating each element in isolation.

The methodological contribution may prove as important as the substantive findings. The authors present their approach as a new modeling tool for analyzing the relative importance of nodes and edges in the scientific literature, and they have made the data collection structure and results available in a supplementary document to ensure reproducibility, including the exact search queries used for each component pair. This transparency means other researchers can update the analysis as new publications appear, extend the network with additional nodes, or apply the same technique to neighboring concepts such as sustainable cities, resilient cities, or low-carbon cities, concepts that the same research group has previously compared and untangled. In a field often criticized for conceptual vagueness and vendor-driven hype, a reproducible, quantitative map of what the smart city actually consists of, and how that composition has changed over a quarter century, is a genuinely useful piece of infrastructure for the research community itself.

Ultimately, the study is a reminder that the smart city is less a product than a process, an evolving web of relationships among data, technology, people, and institutions rather than a fixed blueprint to be purchased and installed. By modeling that web explicitly, the researchers have given urban scholars a common language for debating which connections deserve attention and which assumptions deserve skepticism. As cities around the world continue to digitize, the question is no longer whether data assets sit at the center of urban governance, the evidence now shows they do, but whether the rest of the network, the human and institutional nodes that give data its meaning, will receive the investment they need to keep the whole system connected.

Subject of Research: Network-based modeling of smart city governance components and their evolution in the scientific literature

Article Title: Smart city modeling: a social network analysis approach

Article References: Noori, N., de Jong, M., Joss, S., & Ranjbar-Sahraei, B. (2024). Smart city modeling: a social network analysis approach. Global Public Policy and Governance, 4(4), 420-446. https://doi.org/10.1007/s43508-024-00101-9

Image Credits: AI Generated

DOI: 10.1007/s43508-024-00101-9

Keywords: smart cities, social network analysis, input–output modeling, bibliometrics, data assets, urban governance, smart mobility, human capital, ICT infrastructure, sustainability, network models, public policy

The new study, published in the International Journal of Data Science and Analytics by Akilu Rilwan Muhammad, Ana Aguiar, and João Mendes-Moreira, presents a class-aware classification framework called Class-subspace. Its central innovation lies in borrowing an idea from cooperative game theory: the Shapley value. Originally developed to fairly divide a payout among players in a coalition, Shapley values have become one of the most trusted tools for explaining machine learning predictions through the SHAP method, short for SHapley Additive exPlanations. Rather than using SHAP merely to explain a model after the fact, the Porto team flips the workflow, using SHAP attributions to actively construct the model itself.

The reasoning is deceptively simple. Conventional transportation mode detection pipelines apply a single, globally selected feature set across every travel mode, optimising one subset of variables for the entire multi-class problem. But the researchers found that the features most relevant for distinguishing a bus journey were often sub-optimal for identifying a car trip. A single global feature set, they argue, necessarily compromises performance for certain classes. Class-subspace instead trains an initial classifier, computes mean absolute SHAP values for each feature and each class, ranks the features by their contribution to each mode’s prediction, and retains the minimal set of features that explains a cumulative threshold of 60 percent of each class’s relevance. Each transportation mode thus acquires its own compact feature subspace, and a specialised base learner is trained for every mode before their votes are aggregated into a final decision.

The technical pipeline behind the framework is substantial. The team worked with two real-world GPS trajectory datasets: SenseMyFEUP, collected from 227 participants in Porto, Portugal, using an Android application that sampled location roughly once per second, and the well-known GeoLife dataset from Microsoft Research Asia, gathered from 182 participants in Beijing over nearly five years. Rigorous preprocessing was essential. The researchers chained GPS points into trips using a 30-minute gap threshold, filtered out anomalous sessions that continued recording without movement, applied a spatial filter to keep only city-scale journeys, and capped implausible walking trips at eight kilometres. Segments shorter than 50 metres were discarded, removing nearly 39 percent of segments in the Porto data and about 14 percent in GeoLife.

Feature engineering drew on both the time and frequency domains. For each 100-step instance, the team computed statistics of speed, acceleration, jerk, and bearing rate, capturing the kinematic character of motion. They then applied the Fast Fourier Transform to convert these signals into the frequency domain, hypothesising that cyclic human movements such as walking or pedalling would produce distinctive high-frequency components, while smooth mechanical travel by car or bus would concentrate energy at low frequencies. In total, 80 features per instance were generated, with frequency-domain information encoded through the indices of the top ten spectral components rather than their magnitudes, capturing which frequency bands are active for each mode rather than how much energy they carry.

Evaluation was deliberately conservative. Instead of the random data splits common in the field, which can leak future patterns into training and inflate accuracy, the team used temporal splits, training on earlier data and testing on later periods. Against baselines including Sequential Forward Floating Selection, Mutual Information filtering, and the Boruta algorithm, Class-subspace achieved weighted ROC-AUC scores of up to 78.8 percent on the Porto dataset and 89.5 percent on GeoLife using the XGBoost classifier, performing comparably to established global feature selection methods. A formal statistical comparison using the Friedman test and Bonferroni-Dunn post hoc analysis confirmed that no baseline differed significantly from the proposed method.

What the framework sacrifices in raw accuracy gains, it returns in interpretability. An analysis of feature overlap across class-specific subspaces, measured with the Jaccard similarity index, revealed moderate overlap and a small universal core of speed-based statistics, including maximum, mean, standard deviation, and 85th-percentile speed, that appeared in every configuration. Beyond that shared core, each mode carried its own signature. Walking was captured almost entirely by speed statistics, consistent with its narrow and stable range. Cycling added bearing rate features reflecting frequent directional changes. Bus detection drew on jerk and bearing rate, echoing stop-and-go dynamics. Metro uniquely selected bearing rate distributions and acceleration frequency components, matching the confined, scheduled vibration patterns of rail transit. On the Porto data with random forest, only 7 of 32 distinct features were common to all classes, while 9 served exclusively as metro signatures.

The authors are candid about limitations. Bus and metro classes remained difficult on the Porto dataset, a challenge they attribute to the joint effect of class imbalance and class overlap rather than to feature selection itself, pointing to their earlier work characterising these issues. The computational cost of Shapley values is a known constraint, and the sensitivity of performance to the contribution threshold varied between datasets, with GeoLife remarkably stable across thresholds while the Porto data peaked sharply at 0.60. Future work, they suggest, could explore adaptive threshold selection through cross-validated grid search, Bayesian optimisation, or information-theoretic criteria, and extend the class-subspace concept to other multi-class domains.

The broader implications reach well beyond transportation research. As cities worldwide lean on crowdsensed mobility data for planning, emissions modelling, and infrastructure investment, the trustworthiness of the underlying classification models matters enormously. Class-subspace demonstrates that per-class interpretability need not come at the price of predictive power: a model can match the performance of globally optimised baselines while explicitly revealing which evidence supports each decision. In a field where reported accuracies above 90 percent sometimes rest on leaky evaluation protocols, the Porto team’s insistence on temporal splits and statistical rigor sets a standard. Their message to the machine learning community is clear: when classes are fundamentally different, stop forcing them to share the same lens, and let game theory hand each one its own.

Subject of Research: SHAP-driven class-aware machine learning for transportation mode detection from GPS trajectory data

Article Title: Class-subspace learning: a SHAP-driven framework for class-aware transportation mode detection

Article References: Muhammad, A. R., Aguiar, A., & Mendes-Moreira, J. (2026). Class-subspace learning: a SHAP-driven framework for class-aware transportation mode detection. International Journal of Data Science and Analytics, 22(1), Article 302. https://doi.org/10.1007/s41060-026-01263-x

Image Credits: AI Generated

DOI: 10.1007/s41060-026-01263-x

Keywords: transportation mode detection, SHAP values, machine learning, GPS trajectories, feature selection, Shapley values, urban mobility, smart mobility, interpretability, XGBoost, Fast Fourier Transform, class-aware classification