Bias reduction via cooperative bargaining in synthetic graph dataset generation

In general, to draw robust conclusions from a dataset, all the analyzed population must be represented on said dataset. Having a dataset that does not fulfill this condition normally leads to selection bias. This problem can affect any dataset, including graph datasets, which have become popular wit...

Descripción completa

Detalles Bibliográficos
Autores: Wassington, Axel, Abadal Cavallé, Sergi|||0000-0003-0941-0260
Tipo de recurso: artículo
Fecha de publicación:2025
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/426426
Acceso en línea:https://hdl.handle.net/2117/426426
https://dx.doi.org/10.1007/s10489-024-05947-4
Access Level:acceso abierto
Palabra clave:Graph theory
Random graph generation
Selection bias
Synthetic dataset
Graph neural networks
Graph processing
Àrees temàtiques de la UPC::Informàtica::Intel·ligència artificial
Descripción
Sumario:In general, to draw robust conclusions from a dataset, all the analyzed population must be represented on said dataset. Having a dataset that does not fulfill this condition normally leads to selection bias. This problem can affect any dataset, including graph datasets, which have become popular with the emergence of Graph Neural Networks (GNNs) and their many applications. Although synthetic graphs can be used to augment available real graph datasets to overcome selection bias, the generation of unbiased synthetic datasets is complex with current tools. In this work, we propose a method to find a synthetic graph dataset that has a well-distributed representation of graphs within a given metric space. The resulting dataset can then be used, among others, to study the accuracy of different GNN models or to benchmark the speedups obtained by different graph processing acceleration frameworks.