Evolving Knowledge Graph-Based Knowledge Diffusion Model

Abstract

Knowledge diffusion is capable of narrowing the knowledge gap between individuals and is conducive to transforming knowledge into economically and socially valuable outcomes. Although knowledge diffusion is of particular importance, efficient modeling and analysis of knowledge diffusion may not be achieved easily. In this paper, we apply the concept of evolving knowledge graph (EKG) in modeling and analyzing knowledge diffusion and propose an EKG-based knowledge diffusion model. To characterize the relationships between researchers and papers in the proposed network model, we define three subnetworks, i.e., relationship subnetwork, citation subnetwork and author-paper subnetwork, and analyze the evolution process of the subnetworks. The knowledge diffusion process in the proposed network model is then examined. We analyze respectively the impacts of interpersonal contact and paper citation in knowledge diffusion, and propose a modified susceptible-infective-recovered-susceptible (SIRS) epidemic model-based knowledge diffusion process. Theoretical analysis is conducted to examine the expected values of the weighted degree of network entities, and the transmission threshold of knowledge diffusion process. Extensive simulations are performed to evaluate the performance of the proposed network model.