Complex Network Approach Research Articles

Identification of City Hotspots by Analyzing Telecom Call Detail Records Using Complex Network Modeling

Many interesting observations are made from the analysis of Telecom bigdata called call detail records (CDR) on various aspects of city dynamics. CDR is a low-cost, real-time, and clean (no human error) data with abundant availability to the state administrations. However, due to its massive size, the processing of raw CDR is not feasible most of the time. Graph-theoretic approaches to network analysis are enriched with efficient algorithms and heuristics for modeling and analyzing various generic problems. Once a CDR dataset is modeled as a communication network with aggregation of raw data spatially and temporally, network analysis tools yield significant results. A heightened communication activity (call/SMS/Internet) originating from a location/area to many other areas signify the importance of that originating area. In this paper, we have modeled the telecommunication data of two Italian cities (Milan and Trento) as a weighted network. The modeled network from raw CDR becomes very dense due to the large number of edges. We have pre-processed and thresholded the available CDR bigdata to generate many smaller-sized sparse networks by removing less significant edges. Different well-known node-centrality metrics (weighted degree, weighted k-shell) and our proposed modified edge weight degree neighborhood (EwDN) method are applied on the modeled network to identify the top-k most important nodes representing the hotspot locations in the city. We also use five days data and compare changes in the hotspot patterns. Finally, we validate our findings (hotspot locations) with ground truth from other sources using projected geo-location tagging on QGIS (An open source Geographic Information System application tool). To overcome the availability of limited ground truth, we use the standard Susceptible–Infectious–Recovered (SIR) model as a benchmark and compare various methods using Kendall’s rank correlation. Results indicate that the proposed network modeling, size reduction by edge-weight threshold, and application of complex network approaches including the proposed modified EwDN method identify important places in almost real-time with much less computational overhead.

Analysis of Turkey’s airport network structure and centrality in the opening-out period after the first wave of COVID-19: A complex network approach

There are a few studies present analyzing air transport structures of the countries, regions, and the world by using complex network theory. Although Turkey has a complex air network with 56 airports, thorough research has not been carried out applying complex network analysis to reveal the structure of the airport network in Turkey yet. Furthermore, the fact that Turkey is a developing economy and the role of air transport in this development is undeniable, and its use as a transit area due to its geopolitical location, together with social and political factors, makes it important to understand this air transport network in detail. Consequently, in this paper, Turkey’s Airport Network (TAN) topology was explored by implementing concepts of complex network theory using domestic and international passenger flight data gathered from FlightRadar during the opening period after the first wave of Covid-19. The average path length and clustering coefficient for connectivity performance and centrality metrics (degree, betweenness, and closeness) were computed and the network correlations were also measured and compared with simulated random, small-world (SW), and scale-free (SF) network models. Also, structural similarities and differences with the air networks of other countries are revealed.

Time Series Analysis of Land Cover Change Using Remotely Sensed and Multisource Urban Data Based on Machine Learning: A Case Study of Shenzhen, China from 1979 to 2022

Shenzhen has experienced rapid urbanization since the establishment of the Special Economic Zone in 1978. However, it is rare to witness high-speed urbanization in Shenzhen. It is important to study the LUCC progress in Shenzhen (regarding refusing multisource data), which can provide a reference for governments to solve the problems of land resource shortages and urban expansion spaces. In this paper, nine Landsat images were used to retrieve land cover maps in Shenzhen, China, from 1979 to 2022. The classification method is based on support vector machines with assistance from visual interpretation. The results show that the urban area increased by 756.84 km2, the vegetation area decreased by 546.27 km2, the water area decreased by 132.95 km2, and the bare area decreased by 77.62 km2 in the last 43 years of our research region. Urban sprawl starts from the Luohu district, then propagates to Futian, Nanshan, and Yantian districts, and finally expands to other outlying districts (Baoan, Longgang, Guangming, Dapeng, and Pingshan). The spatial–temporal characteristics and the impact factors of urbanization were further analyzed. The visualization of land cover changes based on a complex network approach reveals that the velocity of urban expansion is growing. The coastline distributions were retrieved from nine observation times from 1979 to 2022; the results show that the west coastline changed more dramatically than the east and most of the east coastline remained stable, except for the parts near Yantian port and Mirs Bay, which experienced some changes. The impact factors of coastline changes are further discussed. Through a correlation analysis using urban data, such as transportation and socioeconomic factors, it was found that elevation and roads have strong constraints on the spatial patterns of a city’s expansion. There is exponential decay in the urban land increase against the distance to the roads, implying that traffic factors greatly determine urban land expansion. The turning point of the exponential decay is a distance of around 150 m. Time and population are highly correlated with land use development, indicating that urban land grows linearly with time and the population, which are important driving forces of urban land development. Compared with secondary and tertiary industries, the primary industry is less related to urban land use in Shenzhen.

Detecting Community Structure in Financial Markets Using the Bat Optimization Algorithm

A lucid representation of the hidden structure of real-world application has attracted complex network research communities and triggered a vast number of solutions in order to resolve complex network issues. In the same direction, initially, this paper proposes a methodology to act on the financial dataset and construct a stock correlation network of four stock indexes based on the closing stock price. The significance of this research work is to form an effective stock community based on their complex price pattern dependencies (i.e., simultaneous fluctuations in stock prices of companies in a time series data). This paper proposes a community detection approach for stock correlation complex networks using the BAT optimization algorithm aiming to achieve high modularity and better-correlated communities. Theoretical analysis and empirical modularity performance measure results have shown that the usage of BAT algorithm for community detection proves to transcend performance in comparison to standard network community detection algorithms – greedy and label propagation.

A linear time series analysis of carbon price via a complex network approach

Identifying the essential characteristics and forecasting carbon prices is significant in promoting green transformation. This study transforms the time series into networks based on China’s pilots by using the visibility graph, mining more information on the structure features. Then, we calculate nodes’ similarity to forecast the carbon prices by link prediction. To improve the predicted accuracy, we notice the node distance to introduce the weight coefficient, measuring the impact of different nodes on future nodes. Finally, this study divides eight pilots into different communities by hierarchical clustering to study the similarities between these pilots. The results show that eight pilots are the “small world” networks except for Chongqing and Shenzhen pilots, all of which are “scale-free” networks except for Shanghai and Tianjin pilots. Compared with other predicted methods, the proposed method in this study has good predicted performance. Moreover, these eight pilots are divided into three clusters, indicating a higher similarity in their price-setting schemes in the same community. Based on the analysis of China’s pilots, this study provides references for carbon trading and related enterprises.

The Evaluation and Key-Factor Identification of the Influence of Tourism on the Soil of Mount Tai

Tourism has been proven to disturb the soil environments and stimulate heavy metal accumulation in scenic areas. Currently, research on the driving mechanisms of the impact of tourism on soil quality degradation is limited. Therefore, the aim of this study was to introduce a complex network method to comprehensively depict the impact of tourism on soil quality. To explore the key influencing factors, we collected 10 topsoil samples and 10 corresponding control samples from representative scenic areas in Mount Tai. Soil physicochemical properties (organic matter (OM), alkali dispelled nitrogen (AN), total nitrogen (TN), available phosphorus (AP), available potassium (AK), electrical conductivity (EC)), heavy metal (Cr, Cd, Pb, Hg, As, Cu) content, and microbial community diversity (by Eco-plate method) were analyzed. Additionally, complex networks of soil quality variables were established based on Pearson correlation coefficients. The results show that the OM, AN, and AP contents of scenic areas are 1.2, 1.03 and 1.18 times higher, while the AK content is 0.97 times lower, than those of the control sites, respectively (mean values of samples from scenic area vs. contorl sites). The single factor index of Hg, As, Pb, Cd, Cu, and Cr are increased from 3.65, 0.53, 0.85, 1.25, 0.78 and 0.58 to 3.69, 0.57, 1.24, 1.75, 0.97, 0.63 and 3.19, respectively, which means that tourism significantly exacerbates soil heavy metal accumulation. Additionally, the soil microbial activity and diversity are slightly reduced due to tourism. In general, the difference in the mean value of each soil quality variable between the scenic spot and the control site is not significant. However, tourism significantly reduced the connectivity and integrity of soil quality variables, which ultimately destabilized the soil, inferred from the comparison of the network’s topological parameters. Therefore, raising the content of soil OM and AN and controlling Cd pollution should be given more priority in soil ecosystem protection to counteract the negative impact of tourism on Mount Tai. It was the major limitation of the study that few typical scenic spots were selected as sampling points on Mount Tai. However, this study is sufficient to show that the complex network approach can be extended to other similar studies of soil quality degradation driving mechanisms.

The carbon effects of the evolution of node status in the world trade network

Trade has contributed to economic development and has brought countries or regions of the world closer together, but it has also had a significant impact on the global environment, particularly in terms of carbon dioxide emissions. Does an increase in a country’s trade necessarily contribute to an increase in its carbon emissions? This paper examines the impact of changes in the position of major countries in the world trade network on their carbon emission levels. In terms of research methodology, this paper achieves an innovation by adopting a complex network approach to analyze the structural characteristics of the trade relationship networks of major countries or regions in the world, providing a new perspective for the study of the relationship between trade development and carbon emissions. The results of the study show that: trade relations among countries are relatively stable from 2000 to 2020, trade ties among members of regional integration organizations are increasing, the top ten countries in terms of importance are mainly developed countries, and China has very close trade relations with most countries. Based on the analysis of structural characteristics, the impact of changes in the network status of each country on its ocarbon emissions is analyzed, using indicators such as the degree centrality of each node as the independent variable and its domestic carbon emission level as the dependent variable. It is found that developed countries have a significant positive impact on in-going degree centrality, and insignificant impact on out-going degree centrality and betweenness centrality. In contrast, developing countries have a significant positive impact on out-going degree centrality and a negative impact on carbon emissions by betweenness centrality, and this conclusion tells us that not all export growth will contribute to higher levels of carbon emissions in the country. Eigenvector centrality has a negative effect on carbon emissions for both developing and developed countries, and closeness to centrality has no effect on carbon emissions while closeness centrality has no effect on carbon emissions. The results of this study again show that the influence of developed countries is greater than that of developing countries on carbon emissions; therefore, the role of developed countries can be taken into account in subsequent studies on carbon emission reduction.

Urban air quality: What is the optimal place to reduce transport emissions?

We develop a linear model based on a complex network approach that predicts the effect of emission changes on air pollution exposure in urban street networks including NO–NO2–O3-chemisty. The operational air quality model SIRANE is used to create a weighted adjacency matrix A describing the relation between emissions of a passive scalar inside streets and the resulting concentrations in the street network. A case study in South Kensington (London) is used, and the adjacency matrix A0 is determined for one wind speed and eight different wind directions. The physics of the underlying problem is used to infer A for different wind speeds. Good agreement between SIRANE predictions and the model is observed for all but the lowest wind speed, despite non-linearities in SIRANE's model formulation. An indicator for exposure in the street is developed, and it is shown that the out-degree of the exposure matrix E represents the effect of a change in emissions on the exposure reduction in all streets in the network. The approach is then extended to NO–NO2–O3-chemisty, which introduces a non-linearity. It is shown that a linearised model agrees well with the fully nonlinear SIRANE predictions. The model shows that roads with large height-to-width ratios are the first in which emissions should be reduced in order to maximise exposure reduction.

Effect and Mechanism of Economic Circulation in the Middle and Lower Reaches of the Yellow River: Multiregional Input–Output Model and Industrial Complex Network Approaches

China has implemented the Yellow River strategy, and the middle and lower reaches of the Yellow River (MLYR) play an important role in promoting the sustainable economic growth of China. However, the economic circulation of the MLYR is constricted by the imbalance and heterogeneity in the economy in the regions, and it is necessary to explore how economic circulation and sustainable development in the MLYR can be improved. In this study, based on the multiregional input–output tables for 2012 and 2017, we developed a MLYR multiregional input–output model; applied indicators, such as intraregional multiplier, interregional feedback, and spillover, to measure economic circulation effects; further developed the industrial circular network; and designed indicators of cycle length distribution, average cycle correlation, influence of the industrial cycle, and interactions of the weighted cycle to analyze the industrial circulation mechanism in the MLYR. We also analyzed the spatial and industrial structures of the economic circulation flows. The results show that economic linkages have been strengthened to a certain extent, but the imbalance in economic circulation is still prominent, and the imbalanced circular effects are determined by the characteristics of the cycles in the MLYR. The empirical findings contribute to several aspects of the theory of imbalanced economic development and provide an important perspective on, and feasible path for, achieving economic development. We suggest that policymakers should build a multi-dimensional innovation cooperation system, improve the digital connectivity of regions, and promote the green and low-carbon development of industry and the application of new energy technologies to achieve balanced, common, and sustainable economic development in the MLYR.

Evidence of self-organized criticality in time series by the horizontal visibility graph approach

Determination of self-organized criticality (SOC) is crucial in evaluating the dynamical behavior of a time series. Here, we apply the complex network approach to assess the SOC characteristics in synthesis and real-world data sets. For this purpose, we employ the horizontal visibility graph (HVG) method and construct the relevant networks for two numerical avalanche-based samples (i.e., sand-pile models), several financial markets, and a solar nano-flare emission model. These series are shown to have long-temporal correlations via the detrended fluctuation analysis. We compute the degree distribution, maximum eigenvalue, and average clustering coefficient of the constructed HVGs and compare them with the values obtained for random and chaotic processes. The results manifest a perceptible deviation between these parameters in random and SOC time series. We conclude that the mentioned HVG’s features can distinguish between SOC and random systems.

A complex network study on Guangdong-Hong Kong-Macao Greater Bay Area based on economic development index

City cluster, the most advanced spatial organization for urban development in the mature stage, is an important carrier for regional economic development. This paper researched into the correlation of economic development changes among cities in the Guangdong-Hong Kong-Macao Greater Bay Area through a new complex network approach. First, Principal Component Analysis was adopted to downscale the economic indicators of each city in the Greater Bay Area, and then we construct an economic development index on this basis; Second, through a TVS-VAR-SV model, an complex network of economic development changes within the city clusters in the Greater Bay Area was built based on the economic development index of each city mentioned above. It can be concluded from the study that Macao, firstly, is the biggest beneficiary of the complex network. Secondly, the most important node in the network is Guangzhou, whose development has a multiply effect on the development of the whole system. Thirdly, Hong Kong and Shenzhen enjoy the highest level of economic development, but their spillover effect on the city cluster system is much lower than that of Guangzhou.

City2vec: Urban knowledge discovery based on population mobile network

Due to the needs of social and economic development, population movements between cities often occur on a large scale. Spontaneous population movements between cities constitute a Mobile Network of enormous scale, and although a considerable amount of research has been conducted to parse this structure using algorithms related to complex networks, it is not possible to quantitatively describe the differences and similarities between different cities. We use graph embedding algorithm that extends the traditional complex network approach of one-dimensional cognition of cities to two-dimensional cognition. It could project city information into a high-dimensional mathematical space by learning the population flow relationships and computing a unique vector representation. We parsed the cell phone number data in about 80 million POIs in 334 cities across China provided by Tencent eMap, and constructed a city Mobile Network containing 2,662,596 directed edges that could effectively capture long-term, long-distance population migrations. The city embedding is actually a reduced dimensional representation of the mobile network, which can retain richer original information. Our research method can not only accomplish the tasks that traditional complex network analysis methods can do well, but also achieve higher dimensional cognition of cities (e.g., spatial relationships between cities and maturity of urban agglomerations), which is an effective complement to traditional graph theory methods. • Capturing long-term population movements based on electronic map phone number data. • Raising the level of awareness of the city. • Compute embedding representations of nodes in city networks. • City vectors contain geographic location and culture attribute.

Multiscale online-horizontal-visibility-graph correlation analysis of financial market

In this study, we introduce an analogous multiscale online-horizontal-visibility-graph correlation analysis (MOHVGCA) method to quantify the association strength between pairwise nonlinear time series. The new method is based on complex network approaches and proved to be a powerful and efficient tool for large-scale time series analysis. The versatility and robustness of this new method are illustrated through synthetic series. Results obtained allow us to be optimistic about its efficiency in correlation analysis of complex time series. Furthermore, we employ the new method to characterize the association between WTI Crude Oil futures and representative traded currencies on 5 min intra-day recordings in the period January 2013–December 2015. The results reveal that there exist positive associations between WTI Crude Oil and the considered traded currencies, and the association degree between WTI Crude Oil and currencies from oil-exporting and oil-importing countries is significantly different over different scales. The oil crisis weakens the association between oil and the currencies from the oil-importing currencies and the information exchange between oil and currencies from oil-exporting countries strengthened dramatically after the oil crisis. Even more, we detect that the Russian ruble always has the strongest ties with oil in different periods. These findings shed new light on the correlation analysis from the perspective of multiscale network analysis.

Similarity-based heterogeneity and cohesiveness of networked companies issuing minibonds

This paper adopts a complex network approach for discussing the level of heterogeneity and cohesiveness among firms that have used a particular financial instrument — the so-called minibond. The nodes of the networks represent firms, and the weight of a link is assumed to be increasing with the similarity of the corresponding nodes/firms — where similarity is intended in terms of specific economic-financial characteristics of the firms. We assess the level of heterogeneity through the strength degree and the level of cohesiveness through the clustering coefficient. The empirical experiments are based on the paradigmatic case of the Italian reality, where minibonds have been and are currently efficiently used. The analysis reveals regularities and discrepancies among firms’ financial characteristics. Furthermore, the results suggest the potential identification of the main determinants of minibonds issuance.

Territorial Development as an Innovation Driver: A Complex Network Approach

Rankings are a well-established tool to evaluate the performance of actors in different sectors of the economy, and their use is increasing even in the context of the startup ecosystem, both on a regional and on a global scale. Although rankings meet the demand for measurability and comparability, they often provide an oversimplified picture of the status quo, which, in particular, overlooks the variability of the socio-economic conditions in which the quantified results are achieved. In this paper, we describe an approach based on constructing a network of world countries, in which links are determined by mutual similarity in terms of development indicators. Through the instrument of community detection, we perform an unsupervised partition of the considered set of countries, aimed at interpreting their performance in the StartupBlink rankings. We consider both the global ranking and the specific ones (quality, quantity, business). After verifying if community membership is predictive of the success of a country in the considered ranking, we rate country performances in terms of the expectation based on community peers. We are thus able to identify cases in which performance is better than expected, providing a benchmark for countries in similar conditions, and cases in which performance is below the expectation, highlighting the need to strengthen the innovation ecosystem.

An evaluation of the international trade-related CO2 emissions for China's light industry sector: A complex network approach

The aim of this paper is to evaluate China's local implied carbon emission (ICE) caused by the international trade of light industrial products. To do that, a multiregional-multisectoral evaluation model was proposed based on complex network theory. We constructed weighted network of China's imports and exports of light industrial products, and calculated the ICE associated with the trade flows. Then an evaluation index system was proposed based on four indicators: in-strength, out-strength, implied CO2 in import, and implied CO2 in export. Next, three multi-attribute decision-making methods and the minimum-deviation method were used to form a multiregional-multisectoral evaluation model, obtaining the level of the ICE coming from trade. We found that the United States, India, the Netherlands, the United Kingdom, and Mexico were the five countries that caused the most serious ICE in China. Within the light industry sector, the most significant light industries responsible for China's local carbon emissions was the household electrical appliances industry. Such results may provide a reference for formulating light industry strategies to have a positive impact on the socio-economic landscape and encourage sustainable development.

Analysis of Dormancy-Associated Transcriptional Networks Reveals a Shared Quiescence Signature in Lung and Colorectal Cancer.

Quiescent cancer cells (QCCs) are a common feature of solid tumors, representing a major obstacle to the long-term success of cancer therapies. We isolated QCCs ex vivo from non-small cell lung cancer (NSCLC) and colorectal cancer (CRC) xenografts with a label-retaining strategy and compared QCCs gene expression profiles to identify a shared “quiescence signature”. Principal Component Analysis (PCA) revealed a specific component neatly discriminating quiescent and replicative phenotypes in NSCLC and CRC. The discriminating component showed significant overlapping, with 688 genes in common including ZEB2, a master regulator of stem cell plasticity and epithelial-to-mesenchymal transition (EMT). Gene set enrichment analysis showed that QCCs of both NSCLC and CRC had an increased expression of factors related to stemness/self renewal, EMT, TGF-β, morphogenesis, cell adhesion and chemotaxis, whereas proliferating cells overexpressed Myc targets and factors involved in RNA metabolism. Eventually, we analyzed in depth by means of a complex network approach, both the ‘morphogenesis module’ and the subset of differentially expressed genes shared by NCSLC and CRC. This allowed us to recognize different gene regulation network wiring for quiescent and proliferating cells and to underpin few genes central for network integration that may represent new therapeutic vulnerabilities. Altogether, our results highlight common regulatory pathways in QCCs of lung and colorectal tumors that may be the target of future therapeutic interventions.

Collective user switching behavior reveals the influence of TV channels and their hidden community structure

Television is the primary medium through which most families access entertainment and information in their daily lives. Thus, understanding users’ TV viewing behavior is meaningful for several practical issues, such as evaluating the influence of TV channels and providing personalized TV recommendations. However, most existing works regarding TV viewing data are limited to basic statistics (e.g., TV ratings). In this paper, we analyze a large-scale TV viewing dataset for a city in China via a complex network approach. We construct a directed network that characterizes the collective channel-switching behavior of viewers. By using the PageRank method, we reveal the influential TV channels that are more in line with people’s expectations than their rankings based on simple TV ratings. We further construct a network in which channels are linked according to their similarity in users’ switching preferences. This network exhibits a clear community structure, which can help TV stations understand which channels are in the bottleneck and which channels have potential. Overall, our work provides a system perspective to evaluate TV channels and their relationships.

Associations among sleep, hematologic profile, and aerobic and anerobic capacity of young swimmers: A complex network approach.

Although the link between sleep and hematological parameters is well-described, it is unclear how this integration affects the swimmer’s performance. The parameters derived from the non-invasive critical velocity protocol have been extensively used to evaluate these athletes, especially the aerobic capacity (critical velocity—CV) and the anaerobic work capacity (AWC). Thus, this study applied the complex network model to verify the influence of sleep and hematological variables on the CV and AWC of young swimmers. Thirty-eight swimmers (male, n = 20; female, n = 18) completed five experimental evaluations. Initially, the athletes attended the laboratory facilities for venous blood collection, anthropometric measurements, and application of sleep questionnaires. Over the 4 subsequent days, athletes performed randomized maximal efforts on distances of 100, 200, 400, and 800-m. The aerobic and anerobic parameters were determined by linear function between distance vs. time, where CV relates to the slope of regression and AWC to y-intercept. Weighted but untargeted networks were generated based on significant (p < 0.05) correlations among variables regardless of the correlation coefficient. Betweenness and eigenvector metrics were used to highlight the more important nodes inside the complex network. Regardless of the centrality metric, basophils and red blood cells appeared as influential nodes in the networks with AWC or CV as targets. The role of other hematologic components was also revealed in these metrics, along with sleep total time. Overall, these results trigger new discussion on the influence of sleep and hematologic profile on the swimmer’s performance, and the relationships presented by this targeted complex network can be an important tool throughout the athlete’s development.

The weighted cross-shareholding complex network: a copula approach to concentration and control in financial markets

In this work, we focus on the cross-shareholding structure in financial markets. Specifically, we build ad hoc indices of concentration and control by employing a complex network approach with a weighted adjacency matrix. To describe their left and right tail dependence properties, we explore the theoretical dependence structure between such indices through copula functions. The theoretical framework has been tested over a high-quality dataset based on the Italian Stock Market. In doing so, we clearly illustrate how the methodological setting works and derive financial insights. In particular, we advance calibration exercises on parametric copulas under the minimization of both Euclidean distance and entropy measure.