Community Detection

Classical network centrality treats nodes as intrinsically meaningful and edges as static conduits, producing distortions in socio-historical or narrative systems. This paper introduces the Network Spatial Credibility (NSC) protocol, an adaptive, multiplex framework built on five axioms, notably structural vacuity (nodes as empty containers) and relational endogeneity (edges as dynamic field outcomes). We model a multiplex gravitational field combining Sarnoff's Law (dyadic gravity), Metcalfe's Law (ambient friction), and a Gresham weighting mechanism that expels statistically anomalous relationships. The protocol separates micro-statistics (Particular Case-raw local gravity) from macro-statistics (General Case-consensus-weighted mass), producing an Edge Surprise Index and a Global Standard Score to detect anomalies, hubs, and lateral agents. Group fragmentation follows Barabási scale-free constraints and a homogeneous approximation of Reed's Law (𝑉 = 2 𝑛) with Christaller crystallization. Crystallized communities dissolve via stochastic social temperature. Edges evolve logistically, conserving energy across layers. The protocol is implemented in Neo4j + GDS and calibrated via MLE.

Le Havre agglomeration (CODAH) includes 16 establishments classified Seveso with high threshold. In the literature, we construct vulnerability maps to help decision makers assess the risk. Such approaches remain static and do take into account the population displacement in the estimation of the vulnerability. We propose a decision making tool based on a dynamic vulnerability map to evaluate the difficulty of evacuation in the different sectors of CODAH. We use a Geographic Information system (GIS) to visualize the map which evolves with the road traffic state through a detection of communities in large graphs algorithm.

Community detection in social networks is highly influenced by the noise, outliers, and choice of clustering parameter tuning methods. Graph autoencoder (GAE) models and their variants have been developed for community detection in the last few years which resolved the noise and clustering parameter tuning issues. However, the performance of these models is not significant. Hence, in this work, a new approach, GAETOPTICS, is presented. It combines the enhanced GAE model and tuned-OPTICS density-based clustering for detecting more accurate communities in the network. GAETOPTICS approach uses three significant advances to improve clustering accuracy: (1) enhancing the GAE architecture to capture deeper networks in complicated systems and avoid overfitting; (2) including a clustering-specific loss to directly optimize the embeddings for clustering and achieve a balance between community preservation and rebuilding; (3) tuning the OPTICS clustering parameters to maintain significant clusters while further lowering noise while using agglomerative clustering to reconstruct the communities. Performance of the proposed model GAETOPTICS is evaluated using normalized mutual information evaluation metrics with respect to the four other baseline algorithms, i.e., Louvain, Label Propagation, GAE with DBSCAN and GAE with KMean. The experimental results show that the GAETOPTICS community detection approach is scalable and data-driven, which makes it appropriate for intricate and large-scale networks.

Geographic adjacency networks encode the pathways along which infectious diseases propagate across administrative boundaries. This study constructs and analyses a district-level adjacency network of India comprising 666 nodes (districts and union territories) and 1,784 edges derived from polygon intersection of the GADM Level-2 boundary shapefile. Four centrality measures (degree, betweenness, eigenvector, and closeness) are computed and correlated with the number of days between India's first reported COVID-19 case (30 January 2020) and each district's first confirmed case, using covid19india.org data spanning April 2020 to June 2021. Louvain community detection identifies 13 mesoscale communities that are geographically coherent but cross state boundaries, suggesting that epidemic spread followed natural geographic corridors rather than administrative lines. All four centrality measures show negative correlations with time-to-first-case (closeness: r =-0.28; degree: r =-0.20; betweenness: r =-0.11; eigenvector: r =-0.06), confirming that structurally central districts were reached sooner. The Siliguri Corridor emerges as the dominant chokepoint. This narrow strip in West Bengal, made up of Darjeeling, Jalpaiguri, Koch Bihar, and Uttar Dinajpur, carries betweenness centrality values 40 to 65% higher than any other region, owing to its unique role as the sole land bridge connecting NorthEast India to the rest of the country. An interactive multiscenario dashboard extends the network analysis to five policy-relevant scenarios: epidemic spread, railway network resilience, healthcare accessibility, disaster relief logistics, and rural connectivity. Github Repo: https://github.com/sanjanasinghx/india-covid-network-analysis

This project report describes the development of a comprehensive framework for understanding the inherent structural resilience within complex evolving networks. Our work focuses on analyzing large-scale trajectory data to identify stable community patterns and model how these structures persist through significant external changes. The paper details the methodological approach for representing dynamic system interactions and characterizing the intrinsic feedback mechanisms that maintain network integrity. This project's findings provide foundational insights for effective system planning and adaptation management.

TUTOR is an application that aims to provide incorporation of entrepreneurial characteristics and self-knowledge to users. Thus, is proposed the suggestion of groups where low competence characteristics must be mitigated by other users in the group. This process is related to Independent Set problem and requires exponential time. In this context, a heuristic algorithm was developed based on GLP heuristic and presented good results, finding best solutions in polynomial time frequently. Despite the small amount of data in TUTOR, the algorithm is able to suggest groups with a few undesirable cases. Conjunto Independente e demanda tempo exponencial. Neste contexto, foi desenvolvido um algoritmo heurístico, baseado na heurística GLP, que apresentou bons resultados, encontrando as melhores soluções conhecidas em tempo polinomial. Apesar do pouco volume de dados no TUTOR, o algoritmo foi capaz de sugerir grupos, apresentado poucos casos indesejáveis. McClelland (1972) mostrou que empreendedores de sucesso possuem algumas características comportamentais que os diferem dos demais. Estas foram organizadas em três grandes grupos, sendo eles: (i) Conjunto de realização; (ii) Conjunto de planejamento; e (iii) Conjunto de poder. Neste contexto, a ferramenta TUTOR, proposta por Pessoa (2011), tem por objetivo monitorar e proporcionar a incorporação de tais características nos usuários, possibilitando a compreensão das mesmas e relacionando-as com atividades cotidianas. Desta forma, a ferramenta proporciona autoconhecimento aos usuários, o qual é a base para o processo de desenvolvimento. Sendo assim, o TUTOR fornece ao usuário a compreensão de características fortes e fracas em seu comportamento, permitindo que o mesmo trabalhe para compensar vulnerabilidades. Este trabalho propõe a ampliação da ferramenta TUTOR, desenvolvendo uma funcionalidade que possibilite a sugestão de equipes de empreendedores com características complementares, de modo a amenizar os pontos fracos de cada

Artificial intelligence (AI) is rapidly reshaping social and institutional life across the European Union (EU), intensifying the need to understand whether AI systems support or undermine human wellbeing. This study maps the scholarly landscape at the intersection of AI, wellbeing, and the EU using a reproducible bibliometric-to-semantic workflow. We retrieved 780 records across Scopus, Web of Science, IEEE Xplore, and OpenAlex (10 December 2025), retained 437 unique abstract-bearing publications after cleaning and deduplication, and embedded abstracts with a transformer-based sentence embedding model. Using cosine similarity, we constructed a semantic similarity network and selected a connectivity–interpretability trade-off threshold (τ = 0.85), producing an active graph of 158 publications linked by 446 edges (global clustering coefficient ≈ 0.49). Louvain community detection revealed a strongly right-skewed cluster-size distribution, with two dominant thematic cores and a long tail of specialized micro-clusters. To interpret clusters, we ranked papers within each community using PageRank, analyzed influential abstracts to assign cluster labels, and synthesized how each community defines and operationalizes wellbeing. Across clusters, wellbeing is consistently framed as a multidimensional construct tied to autonomy, dignity, fairness, privacy, safety, psychological integrity, and trust, with differences emerging by sector (e.g., healthcare), governance frame (EU AI Act/GDPR/NIS2), and methodological orientation (impact assessment versus normative ethics). The findings highlight both consolidation around governance and evaluation agendas and persistent fragmentation in legal and organizational scholarship, pointing to a research gap in integrative work that connects rights-based safeguards to measurable wellbeing outcomes under real-world deployment conditions.

In this paper, we propose a unified framework OCTracker for tracking overlapping community evolution in online social networks. OCTracker adapts a preliminary community structure towards dynamic changes in social networks using a novel density-based approach for detecting overlapping community structures and automatically detects evolutionary events like birth, growth, contraction, merge, split, and death of communities with time. Unlike other density-based community detection methods, the proposed method does not require the neighborhood threshold parameter to be set by the users, rather it automatically determines the same for each node locally.

In this paper, we propose a unified framework, HOCTracker, for tracking the evolution of hierarchical and overlapping communities in online social networks. Unlike most of the dynamic community detection methods, HOCTracker adapts a preliminary community structure towards dynamic changes in social networks using a novel density-based approach for detecting overlapping community structures, and automatically tracks evolutionary events like birth, growth, contraction, merge, split, and death of communities. It uses a novel and efficient log-based approach to map evolutionary relations between communities identified at two consecutive time-steps of a dynamic network, which considerably reduces the number of community comparisons. Moreover, it does not require an ageing function to remove old interactions for identifying community evolutionary events. HOCTracker is applicable to diirected/undirected and weighted/unweighted networks. Experimental results have shown that community structures identified by HOCTracker on some well-known benchmark networks are significant and in general better that the community structures identified by the state-of-the-art methods.

With the advent of Web 2.0/3.0 supported social media, Online Social Networks (OSNs) have emerged as one of the popular communication tools to interact with similar interest groups around the globe. Due to increasing popularity of OSNs and exponential growth in the number of their users, a significant amount of research efforts has been diverted towards analyzing user-generated data available on these networks, and as a result various community mining techniques have been proposed by different research groups. But, most of the existing techniques consider the number of OSN users as a fixed set, which is not always true in a real scenario, rather the OSNs are dynamic in the sense that many users join/leave the network on a regular basis. Considering such dynamism, this chapter presents a density-based community mining method, OCTracker, for tracking overlapping community evolution in online social networks. The proposed approach adapts a preliminary community structure towards dynamic changes in social networks using a novel density-based approach for detecting overlapping community structures and automatically detects evolutionary events including birth, growth, contraction, merge, split, and death of communities with time. Unlike other density-based community detection methods, the proposed method does not require the neighborhood threshold parameter to be set by the users, rather it automatically determines the same for each node locally. Evaluation results on various datasets reveal that the proposed method is computationally efficient and naturally scales to large social networks.

The influence of bacterial communities on the formation of carbonate deposits such as moonmilk was investigated in Altamira Cave (Spain). The study focuses on the relationship between the bacterial communities at moonmilk deposits and those forming white colonizations, which develop sporadically throughout the cave. Using molecular fingerprinting of the metabolically active bacterial communities detected through RNA analyses, the development of white colonizations and moonmilk deposits showed similar bacterial profiles. White colonizations were able to raise the pH as a result of their metabolism (reaching in situ pH values above 8.5) which was proportional to the nutrient supply. Bacterial activity was analyzed by nanorespirometry showing higher metabolic activity from bacterial colonizations than uncolonized areas. Once carbonate deposits were formed, bacterial activity decreased drastically (down to 5.7% of the white colonization activity). This study reports on a specific type of bacterial community leading to moonmilk deposit formation in a cave environment as a result of bacterial metabolism. The consequence of this process is a macroscopic phenomenon of visible carbonate depositions and accumulation in cave environments.

The proliferation of social media has led to the emergence of a new type of geospatial information that defies the conventions of authoritative or volunteered geographic information, yet can be harvested to reveal unique and dynamic information about people and their activities. In this paper we address the identification and mapping of global virtual communities formed around issues of specific national interest. We refer to these connected virtual communities formed around issues related to a specific state as the polycentric virtual equivalent of that state. Identifying, mapping, and analyzing these virtual communities is a novel challenge for our community, and this is the subject we pursue in this paper. We present these communities relative to established conventions of statehood, address the harvesting of relevant geographical information from social media feeds, and discuss the challenge of visualizing such information. In order to do so we use the current geopolitical situation in Syria as a demonstrative example.

The study of the Elizabethan literary landscape has long been paralyzed by an epistemological deadlock: traditional biographical consensus relies on the normalization of sparse commercial records, while heterodox authorship hypotheses rely on untestable rhetorical parallels. This monograph transcends this impasse by shifting the analytical focus from literary style to network topology. Utilizing Neo4j graph database technology, we construct a Dual-Layer Ontological Architecture that strictly bifurcates the Elizabethan intelligence-literary complex into an Evidentiary Layer (archival facts) and an Interpretive Layer (theoretical hypotheses). This methodological rigor resolves the circularity that has historically plagued computational approaches to authorship studies.

UNESCO’s Intangible Cultural Heritage Lists operate as global classificatory infrastructures that shape cultural visibility. Focusing on South America, this study models UNESCO’s primary and secondary descriptors as co-occurrence networks and examines their community structure and representational diversity. Using Louvain modularity and Shannon entropy, we identify thematic concentrations that disproportionately position Andean countries within ritual and devotional clusters, while Brazil and Colombia exhibit higher semantic diversity across communities. Cross-network analysis of associations between primary and secondary descriptors reveals systematic pairings, such as Procession with Religious syncretism, that expose tensions between institutional categories and local epistemologies. These findings provide quantitative evidence of semantic bias embedded in UNESCO’s metadata ontology and show how network-based methods can support critical evaluations of heritage infrastructures. By making the internal structure of UNESCO’s controlled vocabulary visible, this work contributes to debates on cultural representation, metadata governance, and the epistemic consequences of digital heritage classification.

Cookbooks serve as structured records of both ingredient repertoires and the underlying processing logics that define a culture's culinary identity. By modeling five Transylvanian ethnic traditions-Hungarian, Romanian, Transylvanian Saxon, Jewish, and Armenian-as weighted, undirected co-occurrence networks, we found that interethnic connectivity is driven primarily by technological processes rather than simple ingredient presence. Using purposive sampling, we compiled a harmonized corpus of 1409 recipes and applied explicit ingredient normalization (retention, aggregation, and deconstruction) and a 14-class functional taxonomy. We computed density, clustering, modularity, and centrality measures and compared cuisines with a binary Jaccard index, both at the category level and within four course types. Category networks reveal an exceptionally tight Hungarian-Romanian-Armenian triangle (J > 0.95), whereas course-level results show that main dishes exhibit the strongest divergence (J < 0.28). These results support a layered identity model of Transylvanian gastronomy: while shared confectionery frameworks in desserts dissolve ethnic boundaries (M < 0.17), main dishes actively guard cultural boundaries through distinct technological signatures.

We show that the European airspace can be represented as a multi-scale traffic network whose nodes are airports, sectors, or navigation points and links are defined and weighted according to the traffic of flights between the nodes. By using a unique database of the air traffic in the European airspace, we investigate the architecture of these networks with a special emphasis on their community structure. We propose that unsupervised network community detection algorithms can be used to monitor the current use of the airspaces and improve it by guiding the design of new ones. Specifically, we compare the performance of three community detection algorithms, also by using a null model which takes into account the spatial distance between nodes, and we discuss their ability to find communities that could be used to define new control units of the airspace.

Community detection plays a central role in understanding the modular structure of
complex networks. This work introduces Spider Community Detection, a hybrid local–
global algorithm that constructs communities through a depth-bounded geodesic expansion process. Each spider originates from a structurally strong seed node selected using a composite score based on degree, triangle participation, and local clustering. From each seed, the algorithm grows a localized spider-shaped subgraph through bounded breadth-first exploration, where candidate nodes are evaluated using true modularity gain together with a triangle-closure signal. After the initial spider construction, the method applies modularity-guided attachment of the remaining vertices, Louvain-style local refinement, and greedy merge matching under conductance constraints to reconcile local structure with global partition coherence. Experimental evaluations on real and benchmark datasets, including Karate Club, High School, Political Blogs, Cora, and DBLP, show that Spider produces partitions that are competitive with the established methods in terms of ground-truth recovery and structural quality, while yielding communities with sharp boundaries under conductance-sensitive evaluation.

This research uses sophisticated statistical and machine learning techniques to investigate and mitigate the impact of the 5G Corona virus spread misinformation on the Twitter platform using the COVID-19 Misinformation Tweets Labelled Dataset. The research analyzes the temporal distribution characteristics of the spread of misinformation for specific periods in the day, including morning, afternoon, evening, and night, using both categorical and probabilistic approaches. Sentiment analysis using natural language processing (NLP) is done to establish the emotional content analysis of the tweet, while logistic regression, Random Forest, and Naïve Bayes classifiers are used to establish the predictive model for the likelihood that the tweets are either malicious with the help of predictors such as the number of followers, the number of friends, and the hour the tweets are made. The results indicate the highest number of malicious activities takes place late in the night and early morning periods, with the highest levels taking place in the morning periods, with the highest proportionality levels taking place in the morning periods. The results also indicate that the malicious tweets are slightly more negative than their corresponding counterparts in the emotional content analysis results. Of the three classifiers, Random Forest has the highest classification accuracy (AUC = 0.916) in accurately determining the level of the malicious content spread on the Twitter platform, with the highest efficiency in the spread of misinformation on the Twitter platform.

We propose a novel method to find the community structure in complex networks based on an extremal optimization of the value of modularity. The method outperforms the optimal modularity found by the existing algorithms in the literature. We present the results of the algorithm for computer simulated and real networks and compare them with other approaches. The efficiency and accuracy of the method make it feasible to be used for the accurate identification of community structure in large complex networks.

This research uses sophisticated statistical and machine learning techniques to investigate and mitigate the impact of the 5G Corona virus spread misinformation on the Twitter platform using the COVID-19 Misinformation Tweets Labelled Dataset. The research analyzes the temporal distribution characteristics of the spread of misinformation for specific periods in the day, including morning, afternoon, evening, and night, using both categorical and probabilistic approaches. Sentiment analysis using natural language processing (NLP) is done to establish the emotional content analysis of the tweet, while logistic regression, Random Forest, and Naïve Bayes classifiers are used to establish the predictive model for the likelihood that the tweets are either malicious with the help of predictors such as the number of followers, the number of friends, and the hour the tweets are made. The results indicate the highest number of malicious activities takes place late in the night and early morning periods, with the highest levels taking place in the morning periods, with the highest proportionality levels taking place in the morning periods. The results also indicate that the malicious tweets are slightly more negative than their corresponding counterparts in the emotional content analysis results. Of the three classifiers, Random Forest has the highest classification accuracy (AUC = 0.916) in accurately determining the level of the malicious content spread on the Twitter platform, with the highest efficiency in the spread of misinformation on the Twitter platform.

Rational design of biomolecules-proteins, enzymes, antibodies, and small-molecule ligands-requires optimizing multiple competing objectives (binding affinity, specificity, stability, solubility, manufacturability) over combinatorially vast spaces of sequences and three-dimensional conformations. Classical approaches (genetic algorithms, reinforcement learning, bruteforce screening) struggle with multi-objective tradeoffs, highdimensional landscapes, and computational costs. This paper presents a comprehensive hybrid quantum-classical framework that integrates quantum algorithms as global search components within classical design pipelines. We formulate discrete biomolecular design problems as quadratic unconstrained binary optimization (QUBO) or Ising models incorporating approximate energy functions, evolutionary constraints, and machine-learning predictors. These problems are tackled using structured quantum approximate optimization algorithm (QAOA) circuits with problem-specific mixers and constraint encodings (e.g., preserving key charge or hydrophobicity patterns). Quantum-generated candidate sequences, conformations, or ligand scaffolds are then evaluated and refined using physics-based energy minimization and data-driven scoring. We detail explicit encodings for designing protein-binding motifs via single-site saturation mutagenesis and fragment-based ligand optimization. We prove rigorous bounds on QAOA approximation ratios for structured problems, analyze how quantum search complements classical local optimization, and demonstrate via simulations that quantum-assisted candidates exhibit superior diversity and fitness compared to random or classically-optimized baselines. We provide detailed resource estimates for prototype implementations on near-term quantum hardware and identify concrete design problems where quantum advantage is plausible. While full-scale biomolecular design remains beyond current capabilities, this work establishes the algorithmic foundations for quantum-enhanced molecular discovery platforms.

The link prediction problem is an important research area handled in social network analysis. It consists of inferring the potential links to be formed in the futur given a current snapshot of the network. Several methods have been proposed to address this problem but most of them consider it under a certain framework. Yet, social networks data are often incomplete and noisy, thus it is necessary to manage uncertainty in the prediction task. We review in this paper, the link prediction problem under uncertainty using the belief function theory. First, we present a new graph based model for social networks that encapsulates the uncertainties in the link structure. Then, we propose a new approach for link prediction through information fusion of the neighboring nodes.

Much structured data of scientific interest can be represented as networks, where sets of nodes or vertices are joined together in pairs by links or edges. Although these networks may belong to different research areas, there is one property that many of them do have in common: the network community structure. There has been much recent research on identifying communities in networks. However, most existing approaches require complete network information, which is impractical for some networks, e.g. the World Wide Web or the cell phone telecommunication network. Local community detection algorithms have been proposed to solve the problem but their results usually contain many outliers. In this paper, we propose a new measure of local community structure, coupled with a two-phase algorithm that extracts all possible candidates first, and then optimizes the community hierarchy. We also propose a community discovery process for large networks that iteratively finds communities based on our measure. We compare our results with previous methods on real world networks such as the co-purchase network from Amazon. Experimental results verify the feasibility and effectiveness of our approach.