真实世界复杂网络公开数据集

多智能体影响力最大化背景下，复杂网络数据集的构建通常涵盖多个异质性网络结构，以支持算法的性能验证与泛化能力评估。本研究公开了四类不同类型的真实世界复杂网络数据集，旨在验证所提出方法在多样化网络环境中的有效性。这些数据集涵盖多个应用场景，包括科研合作、基础设施、电信网络及在线平台推荐系统，具体描述如下： 1. 合作作者网络：该网络由研究网络科学的科学家构成，节点代表个体科学家，边表示他们之间的合作关系。整个网络包含 379 个节点和 914 条边，适用于分析学术合作模式和知识传播路径。2. 美国电力网网络：该网络描述了美国电网的基础设施拓扑结构，其中 4900 个节点代表电力站，6600 条边表示电力传输线路。此数据集能够用于研究电网的鲁棒性、故障传播以及关键节点识别问题。3. 网页关系网络：该数据集刻画了网页之间的超链接关系，包含 16100 个节点和 25600 条边。网页之间的链接结构对信息传播过程具有重要影响，可用于研究信息流动模式以及网络爬虫的优化策略。4. 亚马逊推荐网络：该数据集描述了亚马逊电商平台上的商品推荐关系，包含 91800 个节点和125700 条边。节点表示商品，边表示用户基于购买或浏览行为建立的推荐链接，该网络适用于研究在线推荐系统的传播特性以及用户行为分析。这些数据集涵盖了多个不同领域的真实世界网络，为复杂网络算法的鲁棒性分析、传播动力学建模以及优化策略验证提供了重要实验基准。通过在不同拓扑结构和传播机制下的测试，该数据集能够帮助研究人员构建更具泛化能力的优化算法，以应对实际应用中的信息传播、关键节点识别和网络控制问题。In the context of multi-agent influence maximization, the construction of complex network datasets typically incorporates multiple heterogeneous network structures to support algorithm performance validation and generalization ability evaluation. This study releases four distinct types of real-world complex network datasets, aiming to validate the effectiveness of the proposed method across diverse network environments. These datasets cover multiple application scenarios, including scientific collaboration, infrastructure, telecommunication networks, and online platform recommendation systems, with detailed descriptions as follows: 1. Co-authorship Network: This network comprises scientists researching network science, where nodes represent individual scientists and edges denote their collaborative relationships. The entire network contains 379 nodes and 914 edges, and is applicable for analyzing academic collaboration patterns and knowledge dissemination paths. 2. U.S. Power Grid Network: This network describes the infrastructure topology of the U.S. power grid, with 4900 nodes representing power stations and 6600 edges indicating power transmission lines. This dataset can be utilized to study grid robustness, fault propagation, and critical node identification problems. 3. Web Link Network: This dataset characterizes the hyperlink relationships between web pages, containing 16100 nodes and 25600 edges. The link structure between web pages exerts a significant influence on information dissemination processes, and can be used to investigate information flow patterns and optimization strategies for web crawlers. 4. Amazon Recommendation Network: This dataset describes the product recommendation relationships on the Amazon e-commerce platform, with 91800 nodes and 125700 edges. Nodes represent products, and edges denote recommendation links established by users based on purchase or browsing behaviors. This network is suitable for studying the dissemination characteristics of online recommendation systems and user behavior analysis. These datasets cover real-world networks across multiple distinct fields, providing critical experimental benchmarks for robustness analysis of complex network algorithms, dissemination dynamics modeling, and optimization strategy validation. Through testing under different topological structures and dissemination mechanisms, these datasets can assist researchers in developing optimization algorithms with stronger generalization capabilities to address information dissemination, critical node identification, and network control issues in practical applications.

搜集汇总

背景与挑战

背景概述

该数据集公开了四类真实世界复杂网络，包括合作作者网络、美国电力网网络、网页关系网络和亚马逊推荐网络，覆盖科研合作、基础设施、电信网络及在线平台推荐系统等多个应用场景。它旨在为复杂网络算法的性能验证、泛化能力评估以及传播动力学建模提供实验基准。