Graph Processing Research Articles

The hitting time of clique factors

AbstractIn [Trans. Am. Math. Soc. 375 (2022), no. 1, 627–668], Kahn gave the strongest possible, affirmative, answer to Shamir's problem, which had been open since the late 1970s: Let and let be divisible by . Then, in the random ‐uniform hypergraph process on vertices, as soon as the last isolated vertex disappears, a perfect matching emerges. In the present work, we prove the analogue of this result for clique factors in the random graph process: at the time that the last vertex joins a copy of the complete graph , the random graph process contains a ‐factor. Our proof draws on a novel sequence of couplings which embeds the random hypergraph process into the cliques of the random graph process. An analogous result is proved for clique factors in the ‐uniform hypergraph process ().

Optimal Transport with Cyclic Symmetry

We propose novel fast algorithms for optimal transport (OT) utilizing a cyclic symmetry structure of input data. Such OT with cyclic symmetry appears universally in various real-world examples: image processing, urban planning, and graph processing. Our main idea is to reduce OT to a small optimization problem that has significantly fewer variables by utilizing cyclic symmetry and various optimization techniques. On the basis of this reduction, our algorithms solve the small optimization problem instead of the original OT. As a result, our algorithms obtain the optimal solution and the objective function value of the original OT faster than solving the original OT directly. In this paper, our focus is on two crucial OT formulations: the linear programming OT (LOT) and the strongly convex-regularized OT, which includes the well-known entropy-regularized OT (EROT). Experiments show the effectiveness of our algorithms for LOT and EROT in synthetic/real-world data that has a strict/approximate cyclic symmetry structure. Through theoretical and experimental results, this paper successfully introduces the concept of symmetry into the OT research field for the first time.

GraphScale: Scalable Processing on FPGAs for HBM and Large Graphs

Recent advances in graph processing on FPGAs promise to alleviate performance bottlenecks with irregular memory access patterns. Such bottlenecks challenge performance for a growing number of important application areas like machine learning and data analytics. While FPGAs denote a promising solution through flexible memory hierarchies and massive parallelism, we argue that current graph processing accelerators either use the off-chip memory bandwidth inefficiently or do not scale well across memory channels. In this work, we propose GraphScale, a scalable graph processing framework for FPGAs. GraphScale combines multi-channel memory with asynchronous graph processing (i.e., for fast convergence on results) and a compressed graph representation (i.e., for efficient usage of memory bandwidth and reduced memory footprint). GraphScale solves common graph problems like breadth-first search, PageRank, and weakly connected components through modular user-defined functions, a novel two-dimensional partitioning scheme, and a high-performance two-level crossbar design. Additionally, we extend GraphScale to scale to modern high-bandwidth memory (HBM) and reduce partitioning overhead of large graphs with binary packing.

Optimizing Nested Recursive Queries

Datalog is a declarative programming language that has gained popularity in various domains due to its simplicity, expressiveness, and efficiency. But "pure" Datalog is limited to monotone queries, and cannot be used in most practical applications. For that reason, newer systems are relaxing the language by allowing non-monotone queries to be freely combined with recursion. But by departing from the elegant fixpoint semantics of pure datalog, these systems often result in inefficient query execution, for example they perform redundant computations, or use redundant storage. In this paper, we propose Temporel, a system that allows recursion to be freely combined with non-monotone operators. Temporel optimizes the program by compiling it into a novel intermediate representation that we call TempoDL. Our experimental results show that our system outperforms a state-of-the-art Datalog engine as well as a vectorized and a compiled in-memory database system for a wide range of applications from machine learning to graph processing.

Spatial-Pooling-Based Graph Attention U-Net for Hyperspectral Image Classification

In recent years, graph convolutional networks (GCNs) have attracted increasing attention in hyperspectral image (HSI) classification owing to their exceptional representation capabilities. However, the high computational requirements of GCNs have led most existing GCN-based HSI classification methods to utilize superpixels as graph nodes, thereby limiting the spatial topology scale and neglecting pixel-level spectral–spatial features. To address these limitations, we propose a novel HSI classification network based on graph convolution called the spatial-pooling-based graph attention U-net (SPGAU). Specifically, unlike existing GCN models that rely on fixed graphs, our model involves a spatial pooling method that emulates the region-growing process of superpixels and constructs multi-level graphs by progressively merging adjacent graph nodes. Inspired by the CNN classification framework U-net, SPGAU’s model has a U-shaped structure, realizing multi-scale feature extraction from coarse to fine and gradually fusing features from different graph levels. Additionally, the proposed graph attention convolution method adaptively aggregates adjacency information, thereby further enhancing feature extraction efficiency. Moreover, a 1D-CNN is established to extract pixel-level features, striking an optimal balance between enhancing the feature quality and reducing the computational burden. Experimental results on three representative benchmark datasets demonstrate that the proposed SPGAU outperforms other mainstream models both qualitatively and quantitatively.

Higher rogue and rogue-soliton interaction solutions of a (2 + 1) dimensional nonlinear model in fluid mechanics

In this study, two new theorems are generalized. We obtained a new paradigm about the second order rogue wave and multiple exponential functions, and a new paradigm about the second order rogue wave and multiple hyperbolic cosine functions. Six sets of interaction solutions of the model are solved by means of symbolic calculation and two new theorems. Meaningful graphs of the propagation processes along time demonstrated the interaction phenomena for these solutions. The energy transfer process can be observed when the second order rogue waves interact with multiple exponential functions or multiple hyperbolic cosine functions. As a conclusion from our paper, the solitons’ energy transfers to the second order rogue wave at beginning, the rogue wave’s energy dissipates and transfers to the solitons along the time moving. It will contribute to the research on the generation of rogue waves.

From Dewey to Deep Learning: Exploring the Intellectual Renaissance of Libraries through Artificial Intelligence

Libraries are embracing the potential of Artificial Intelligence (AI) to enhance their services and provide more efficient and personalized experiences to users. This paper explores the role of AI in library services, focusing on its applications and impact. The present article begins by discussing the integration of AI technologies such as natural language processing, machine learning, and knowledge graphs in library systems. It then examines the benefits of AI, including improved information retrieval, recommendation systems, virtual assistants, and data analytics. Ethical considerations related to AI in libraries are also addressed. The paper highlights the challenges and future directions for AI implementation, including the need for training of librarians and the importance of user acceptance. The paper contributes to a better understanding of the opportunities and challenges associated with leveraging AI in library services, ultimately paving the way for more effective and user-centric library experiences.

Towards High-Performance Graph Processing: From a Hardware/Software Co-Design Perspective

Towards High-Performance Graph Processing: From a Hardware/Software Co-Design Perspective

Weisfeiler–Lehman goes dynamic: An analysis of the expressive power of Graph Neural Networks for attributed and dynamic graphs

Graph Neural Networks (GNNs) are a large class of relational models for graph processing. Recent theoretical studies on the expressive power of GNNs have focused on two issues. On the one hand, it has been proven that GNNs are as powerful as the Weisfeiler–Lehman test (1-WL) in their ability to distinguish graphs. Moreover, it has been shown that the equivalence enforced by 1-WL equals unfolding equivalence. On the other hand, GNNs turned out to be universal approximators on graphs modulo the constraints enforced by 1-WL/unfolding equivalence. However, these results only apply to Static Attributed Undirected Homogeneous Graphs (SAUHG) with node attributes. In contrast, real-life applications often involve a much larger variety of graph types. In this paper, we conduct a theoretical analysis of the expressive power of GNNs for two other graph domains that are particularly interesting in practical applications, namely dynamic graphs and SAUGHs with edge attributes. Dynamic graphs are widely used in modern applications; hence, the study of the expressive capability of GNNs in this domain is essential for practical reasons and, in addition, it requires a new analyzing approach due to the difference in the architecture of dynamic GNNs compared to static ones. On the other hand, the examination of SAUHGs is of particular relevance since they act as a standard form for all graph types: it has been shown that all graph types can be transformed without loss of information to SAUHGs with both attributes on nodes and edges. This paper considers generic GNN models and appropriate 1-WL tests for those domains. Then, the known results on the expressive power of GNNs are extended to the mentioned domains: it is proven that GNNs have the same capability as the 1-WL test, the 1-WL equivalence equals unfolding equivalence and that GNNs are universal approximators modulo 1-WL/unfolding equivalence. Moreover, the proof of the approximation capability is mostly constructive and allows us to deduce hints on the architecture of GNNs that can achieve the desired approximation.

Problem-solving activity during the foreign language writing process: A proposal for categorisation and visualisation of source use and a new take on fluency in multilingual writing

Writing processes constitute a complex interplay of planning, formulation and revision. Ideas take shape through the activation of previous knowledge and, when permitted, also its synthesis with information from sources that help to complement it and resolve doubts and shortcomings arising during writing. The possibility to use external help can be especially useful to those writing in a foreign language, and questions about the nature of the sources consulted can contribute new insights into language processing in the multilingual mind, as well as expand our notion of fluency. While leaving the target text is often considered a distraction, a ‘breakdown’ in fluency, it is, in fact, a part of language processing and text creation. This article proposes a novel way to use keylogging data from Inputlog (Leijten and Van Waes 2013) to visualise the crosslinguistic nature of solving language and content problems in L3+ writing: creating process graphs to display the temporal dynamics of different types of sources used. The example data comes from a university-level course on Spanish linguistics, where Spanish was a third or subsequent language for the participants. Evidently, the vast majority of their external activity was language-related and brief, and, interestingly, a great part of it recurs to a lingua franca, English. Some social context and reasoning is offered to explain such an observation.

Web Mining of Online Resources for German Labor Market Research and Education: Finding the Ground Truth?

The labor market is highly dependent on vocational and academic education, training, retraining, and further education in order to master challenges such as advancing digitalization and sustainability. Further training is a key factor in ensuring a qualified workforce, the employability of all employees, and, thus, national competitiveness and innovation. In the contribution at hand, we explore an innovative way to derive knowledge about learning pathways by connecting the dots from different data sources of the German labor market. In particular, we focus on the web mining of online resources for German labor market research and education, such as online advertisements, information portals, and official government websites. A key question for working with different data sources is how to find the ground truth and common data structures that can be used to make the data interoperable. We discuss how to classify and summarize web data from different platforms and which methods can be used for extracting data, entities and relationships from online resources on the German labor market to build a network of educational pathways. Our proposed solution is based on the classification of occupations (KldB) and related document codes (DKZ), and combines natural language processing and knowledge graph technologies. Our research provides the foundation for further investigation into educational pathways and linked data for labor market research. While our work focuses on German data, it is also useful for other German-speaking countries and could easily be extended to other languages such as English.

Disinformation detection using graph neural networks: a survey

The creation and propagation of disinformation on social media is a growing concern. The widespread dissemination of disinformation can have destructive effects on people’s attitudes and behavior. So, it is essential to detect disinformation as soon as possible. Therefore, the interest in effective detection techniques has grown rapidly in recent years. Major social media and social networking sites are trying to develop robust strategies to detect disinformation and prevent its spread. Machine learning techniques and especially neural networks, have an essential role in this task. In this paper, we review different approaches for automatic disinformation detection, with a focus on methods that leverage graph neural networks (GNNs). GNNs are very suitable tools for detecting disinformation in social networks. Because on the one hand, graphs are the most comprehensive way to model social networks and on the other hand, GNNs are the best tool for processing graph data. We define different forms of disinformation, and examine the features used and the methods presented from different perspectives. We also discuss relevant research areas, open problems, and future research directions for disinformation detection in social media.

Shapley-Shubik Agents Within Superstructure-Based Recycling Model: Circular Economy Approaches for Fish Waste Eco-Industrial Park

The tremendous production of fish has resulted in an increased fish waste generation, which ultimately led to the current triple planetary crises on climate, biodiversity, and pollution. In this study, a Fish Waste-based Eco-Industrial Park (FWEIP) model is developed in an attempt to convert the linear economy in existing fish waste management into a circular economy model. Process Graph (P-graph) is used for combinatorial optimization to synthesize optimal FWEIP with the consideration of economic and environmental aspects. The model favors the production of biofuel using the gasification process (Rank 1) with a promising economic benefit of $2.28 million/y without proposing circular synergy within the FWEIP ecosystem. On the other hand, suboptimal solutions—suboptimal 1 (black soldier fly (BSF)) and suboptimal 2 (pyrolysis and gasification) solutions—exhibit gross profit of 17.98% and 24.12% lower than that of the optimal solution. Both suboptimal solutions offer greater circularity with self-sustaining resources (e.g., fish feed, chitosan, and energy). The sensitivity analysis indicates the potential debottlenecking of suboptimal 2 with the use of a catalyst to improve the conversion of bio-oil in the pyrolysis pathway and exhibits a gross profit of 22.54% higher than that of the optimal solution. Following the Shapley-Shubik power index analysis, the hydroponics facility is identified as the pivotal player for both optimal and suboptimal 2 cases with the exception of suboptimal 1 indicating both BSF and hydroponics as a pivotal player. In brief, this research provides the fish waste-based industry with insights and strategies for the implementation of a circular economy as a step toward sustainable development.

Exploring the Perspectives of Patients Living With Lupus: Retrospective Social Listening Study.

Systemic lupus erythematosus (SLE) is a chronic autoimmune inflammatory disease affecting various organs with a wide range of clinical manifestations. Cutaneous lupus erythematosus (CLE) can manifest as a feature of SLE or an independent skin ailment. Health-related quality of life (HRQoL) is frequently compromised in individuals living with lupus. Understanding patients' perspectives when living with a disease is crucial for effectively meeting their unmet needs. Social listening is a promising new method that can provide insights into the experiences of patients living with their disease (lupus) and leverage these insights to inform drug development strategies for addressing their unmet needs. The objective of this study is to explore the experience of patients living with SLE and CLE, including their disease and treatment experiences, HRQoL, and unmet needs, as discussed in web-based social media platforms such as blogs and forums. A retrospective exploratory social listening study was conducted across 13 publicly available English-language social media platforms from October 2019 to January 2022. Data were processed using natural language processing and knowledge graph tagging technology to clean, format, anonymize, and annotate them algorithmically before feeding them to Pharos, a Semalytix proprietary data visualization and analysis platform, for further analysis. Pharos was used to generate descriptive data statistics, providing insights into the magnitude of individual patient experience variables, their differences in the magnitude of variables, and the associations between algorithmically tagged variables. A total of 45,554 posts from 3834 individuals who were algorithmically identified as patients with lupus were included in this study. Among them, 1925 (authoring 5636 posts) and 106 (authoring 243 posts) patients were identified as having SLE and CLE, respectively. Patients frequently mentioned various symptoms in relation to SLE and CLE including pain, fatigue, and rashes; pain and fatigue were identified as the main drivers of HRQoL impairment. The most affected aspects of HRQoL included "mobility," "cognitive capabilities," "recreation and leisure," and "sleep and rest." Existing pharmacological interventions poorly managed the most burdensome symptoms of lupus. Conversely, nonpharmacological treatments, such as exercise and meditation, were frequently associated with HRQoL improvement. Patients with lupus reported a complex interplay of symptoms and HRQoL aspects that negatively influenced one another. This study demonstrates that social listening is an effective method to gather insights into patients' experiences, preferences, and unmet needs, which can be considered during the drug development process to develop effective therapies and improve disease management.

CGgraph: An Ultra-Fast Graph Processing System on Modern Commodity CPU-GPU Co-processor

In recent years, many CPU-GPU heterogeneous graph processing systems have been developed in both academic and industrial to facilitate large-scale graph processing in various applications, e.g., social networks and biological networks. However, the performance of existing systems can be significantly improved by addressing two prevailing challenges: GPU memory over-subscription and efficient CPU-GPU cooperative processing. In this work, we propose CGgraph, an ultra-fast CPU-GPU graph processing system to address these challenges. In particular, CGgraph overcomes GPU-memory over-subscription by extracting a subgraph which only needs to be loaded into GPU memory once, but its vertices and edges can be used in multiple iterations during the graph processing procedure. To support efficient CPU-GPU co-processing, we design a CPU-GPU cooperative processing scheme, which balances the workloads between CPU and GPU by on-demand task allocation. To evaluate the efficiency of CG-graph, we conduct extensive experiments, comparing it with 7 state-of-the-art systems using 4 well-known graph algorithms on 6 real-world graphs. Our prototype system CGgraph outperforms all existing systems, delivering up to an order of magnitude improvement. Moreover, CGgraph on a modern commodity machine with a CPU-GPU co-processor yields superior (or at the very least, comparable) performance compared to existing systems on a high-end CPU-GPU server.

Telematics and diagnosis system for autonomous cargo vehicle

The article deals with the problem of managing the operation and maintenance of autonomous trucks KAMAZ using a telematics and diagnostics system. The functions and composition of the onboard module of a robotic vehicle are described. The main controlled values are determined. The presence of a developed system for monitoring the state of units and assemblies provides a solution to the problem of controlling the modes of the car operation. The structure of the telematics and diagnostics system has been developed and implemented. The main blocks of the proposed system are shown. The main diagnostic and forecasting functions are performed on a workstation with an analytical module. It is proposed to use an approach based on previous conditions matrixes. An example of matrix analysis for a car engine cooling system is given. Modeling of the engine cooling system was carried out and graphs of transient processes were obtained. The analytical module is designed using artificial neural networks to analyze time series of car parameter values. To conduct virtual tests of a car in various operating modes, a simulation model was developed on a stochastic time colored Petri net. The model simulates both the movement of the vehicle to the point of production operations and the change in the technical condition of the car. The model makes it possible to describe and study the influence of random factors on the time of execution of production tasks, to take into account the probabilistic characteristics of failure events or defects in vehicle components and assemblies. The use of an artificial neural network in the analytical module of the diagnostic system workstation makes it possible to predict the technical condition of vehicle components and systems in real time, followed by verification of the dynamics of processes on a simulation model on a Petri net.

A disk I/O optimized system for concurrent graph processing jobs

A disk I/O optimized system for concurrent graph processing jobs

Solution of a Scalar Two-Dimensional Nonlinear Diffraction Problem for Objects of Arbitrary Shape

In this study, the development, design, and software implementation of the methods for solving the nonlinear diffraction problem were performed. The influence of nonlinear medium defined by the Kerr law on the propagation of a wave passing through an object was examined. The differential and integral formulations of the problem and the nonlinear integral equation were considered. The problem was solved for different bodies with the use of various computational grids. Convergence graphs of the iterative processes were generated. The obtained graphical results were presented. The explicit and implicit methods for solving the integral equation were compared.

Power of $k$ Choices in the Semi-Random Graph Process

The semi-random graph process is a single player game in which the player is initially presented an empty graph on $n$ vertices. In each round, a vertex $u$ is presented to the player independently and uniformly at random. The player then adaptively selects a vertex $v$, and adds the edge $uv$ to the graph. For a fixed monotone graph property, the objective of the player is to force the graph to satisfy this property with high probability in as few rounds as possible.
 In this paper, we introduce a natural generalization of this game in which $k$ random vertices $u_1, \ldots, u_k$ are presented to the player in each round. She needs to select one of the presented vertices and connect to any vertex she wants. We focus on the following three monotone properties: minimum degree at least $\ell$, the existence of a perfect matching, and the existence of a Hamiltonian cycle.

Model for describing mivar networks in the format of bipartite and tripartite oriented graphs for decision-making and information processing in machine-building AI

Nowadays, there is a strong trend towards the creation of autonomous systems. For their development, it is required to make decision-making systems, which in this article are proposed to be implemented using mivar technologies of logical artificial intelligence. The paper considers the problem of choosing a specific implementation of the mivar approach using the example of bipartite and threadbare oriented graphs of mivar networks. Purpose – to give an example of activation of parameters and rules when solving the problem of route planning in a two-dimensional logical space. To consider the features of the subject area of automated process control systems and analyze three-sided graphs for decision-making and information processing in mechanical engineering. New ways of developing mivar networks based on three-sided and multi-sided graphs are considered for the implementation of the rules for choosing the type "if ..., then ..., else ...", in which some parameters can be filled in when the condition is met, and others when the condition is not met. The approach described in the paper to the creation of decision-making systems based on mivar technologies allows to achieve a higher degree of autonomy in order to further replace the role of a human operator in automated process control systems in production.