SYNAPSE-4C (SYNthetic Academic Professional Social Ecosystem - 4 Connections)

This dataset is a synthetic academic multiplex social network consisting of approximately 3000 researcher nodes represented across multiple relational layers, including co-authorship, project collaboration, grant collaboration, and academic interaction. The dataset includes interconnected files containing researcher profiles, publication records, project details, grant information, and academic interaction events. The dataset incorporates academic realism through role-based constraints, domain-aware collaboration, temporal consistency, weighted relationships, interdisciplinarity scores, topic vectors, and heterogeneous researcher participation. It is suitable for community detection, overlapping community detection, multiplex fusion, link prediction, graph-based deep learning, collaborator recommendation, and interdisciplinary academic network analysis. Description and Usage of SYNAPSE-4C: 1. Fully synthetic - no real personal data, privacy-safe for open sharing and benchmarking 2. Researcher nodes are shared identically across all four layers, enabling true multiplex analysis 3. Four distinct relational layers capture different collaboration mechanisms: co-authorship, project, grant, and interaction 4. Six clean tabular data files - profiles, temporal evolutions, publications, projects, grants, interactions - each independently usable or joinable 5. Edge weights encode relationship strength or frequency, not just binary presence or absence 6. Temporal fields on every relationship support snapshot analysis, time-slicing, and dynamic graph modeling 7. Topic vectors per researcher node and explicit interdisciplinarity scores enable semantic and cross-domain analysis 8. Role-based and domain-aware generation rules produce graph behavior that mirrors real academic networks 9. Isolated and low-activity nodes are deliberately kept, preserving natural sparsity and making benchmarks more realistic 10.Purpose-built for community detection, link prediction, multiplex layer fusion, and graph neural network experiments本数据集为合成型多层级学术社交网络（academic multiplex social network），包含约3000个研究者节点，涵盖四类关联维度：合著协作、项目合作、基金合作与学术互动。数据集包含多组关联文件，涵盖研究者档案、学术成果发表记录、项目详情、基金资助信息与学术互动事件。该数据集通过基于角色的约束规则、领域感知协作机制、时序一致性约束、加权关系设置、跨学科评分体系、主题向量以及异构研究者参与模式，实现了学术场景的真实性还原。本数据集适用于社区发现（community detection）、重叠社区发现（overlapping community detection）、多层级网络融合（multiplex fusion）、链路预测（link prediction）、基于图的深度学习（graph-based deep learning）、合作者推荐（collaborator recommendation）以及跨学科学术网络分析（interdisciplinary academic network analysis）等研究任务。 SYNAPSE-4C 数据集说明与使用指南： 1. 完全合成生成——不包含任何真实个人数据，可安全开放共享并用作基准测试数据集 2. 所有四层网络均使用完全一致的研究者节点集，可支持真正的多层级网络分析 3. 四层独立关联维度分别捕获不同的协作机制：合著、项目、基金与学术互动 4. 包含6个规范的表格数据文件：研究者档案、时序演化数据、学术成果、项目、基金与互动记录，各文件既可独立使用，也可关联联用 5. 边权重编码了关系强度或交互频率，而非仅使用二元的存在/不存在标记 6. 所有关系均带有时序字段，可支持快照分析、时间切片建模与动态图构建 7. 每个研究者节点均配有主题向量，并提供明确的跨学科评分，可支持语义分析与跨领域研究 8. 基于角色与领域感知的生成规则，可生成贴合真实学术网络特征的图结构行为 9. 刻意保留了孤立节点与低活跃度节点，以还原自然网络的稀疏性特征，使基准测试更贴近真实场景 10. 专为社区发现（community detection）、链路预测（link prediction）、多层级网络层融合（multiplex layer fusion）以及图神经网络（graph neural network）实验打造