A distributed hypergraph model for the full-scale simulation of the collaborations in dblp

TL;DR

This paper introduces a distributed hypergraph model that simulates large-scale collaboration dynamics in dblp, capturing evolving patterns and providing insights into self-organized systems driven by cooperative games.

Contribution

It presents a novel distributed hypergraph model based on Lotka's law and cooperative game theory for simulating large-scale collaboration networks.

Findings

Successfully reproduces collaboration pattern multimodality.

Accurately models evolution of degree, hyperdegree, and clustering.

Captures the growth of the giant component over thirty years.

Abstract

This study proposed a model to give a full-scale simulation for the dynamics of the collaborations in the dblp dataset. It is a distributed model with the capability of simulating large hypergraphs, namely systems with heterogeneously multinary relationship. Its assembly mechanism of hyperedges is driven by Lotka's law and a cooperative game that maximizes benefit-cost ratio for collaborations. The model is built on a circle to express the game, expressing the cost by the distance between nodes. The benefit of coauthoring with a productive researcher or one with many coauthors is expressed by the cumulative degree or hyperdegree of nodes. The model successfully captures the multimodality of collaboration patterns emerged in the dblp dataset, and reproduces the evolutionary trends of collaboration pattern, degree, hyperdegree, clustering, and giant component over thirty years remarkably well. This model has the potential to be extended to understand the complexity of self-organized systems that evolve mainly driven by specific cooperative games, and would be capable of predicting the behavior patterns of system nodes.