方法对比

	指数随机图模型（ERGM / p*）	DBSCAN	图注意力网络
领域	网络分析	机器学习	深度学习
方法族	Process / pipeline	Machine learning	Machine learning
起源年份	1986 (foundational); modern ERGM framework 1996–2007	1996	2018
提出者	Frank & Strauss (1986); extended by Wasserman & Pattison (1996) and Robins et al. (2007)	Ester, M., Kriegel, H.-P., Sander, J. & Xu, X.	Veličković, P. et al.
类型	Probabilistic generative network model	Density-based clustering algorithm	Graph neural network (attention-based)
开创性文献	Robins, G., Pattison, P., Kalish, Y., & Lusher, D. (2007). An introduction to exponential random graph (p*) models for social networks. Social Networks, 29(2), 173-191.	Ester, M., Kriegel, H.-P., Sander, J. & Xu, X. (1996). A Density-Based Algorithm for Discovering Clusters in Large Spatial Databases with Noise. Proceedings of the 2nd KDD, 226–231.	Veličković, P. et al. (2018). Graph Attention Networks. ICLR.
别名	ERGM, p-star model, p* model, Üstel Rastgele Graf Modeli (ERGM / p*)	DBSCAN Kümeleme, density-based clustering, density-based spatial clustering	Graf Dikkat Ağı (GAT), GAT, graph attention network, attention-based graph neural network
相关	6	3	4
摘要	The Exponential Random Graph Model (ERGM), also known as the p* model, is a statistical framework for network analysis that models the probability of an observed network as a function of its local structural features — such as reciprocity, triangles, and degree distribution. Developed from the foundational work of Frank and Strauss (1986) and extended into the modern framework by Wasserman and Pattison (1996) and Robins et al. (2007), ERGM is the inferential standard for social network analysis, capable of testing whether observed network structures arise by chance or reflect genuine social processes.	DBSCAN is a density-based clustering algorithm, introduced by Ester, Kriegel, Sander and Xu in 1996, that groups together points lying in dense regions and flags points in sparse regions as noise. It is effective on noisy data and on clusters of irregular, non-spherical shapes.	The Graph Attention Network (GAT), introduced by Veličković and colleagues in 2018, is a graph neural network variant that learns how much importance to assign to each neighbouring node through a self-attention mechanism. On heterogeneous neighbourhoods and relational classification it produces results superior to graph convolutional networks (GCN).