Improved knowledge diffusion model based on the collaboration hypernetwork

• We present an improved knowledge diffusion hypernetwork model where the knowledge will spread from the target node to all its neighbors.
• The diffusion speed of the presented model is 3.64 times faster than that of traditional knowledge diffusion model.
• It is three times faster to diffuse knowledge by randomly selecting “expert” nodes than that by selecting large-hyperdegree nodes as expert nodes.

The process for absorbing knowledge becomes an essential element for innovation in firms and in adapting to changes in the competitive environment. In this paper, we present an improved knowledge diffusion hypernetwork (IKDH) model based on the idea that knowledge will spread from the target node to all its neighbors in terms of the hyperedge and knowledge stock. We apply the average knowledge stock V(t)V(t), the variable σ2(t)σ2(t), and the variance coefficient c(t)c(t) to evaluate the performance of knowledge diffusion. By analyzing different knowledge diffusion ways, selection ways of the highly knowledgeable nodes, hypernetwork sizes and hypernetwork structures for the performance of knowledge diffusion, results show that the diffusion speed of IKDH model is 3.64 times faster than that of traditional knowledge diffusion (TKDH) model. Besides, it is three times faster to diffuse knowledge by randomly selecting “expert” nodes than that by selecting large-hyperdegree nodes as “expert” nodes. Furthermore, either the closer network structure or smaller network size results in the faster knowledge diffusion.