Graph Theoretic Analysis of Knowledge Networks

TL;DR

This study uses graph theory to analyze 30 real-world Knowledge Networks, revealing their structural weaknesses, limited information sharing, and potential for targeted improvements to enhance communication and coordination.

Contribution

First comparative graph-theoretic analysis of real Knowledge Networks, providing insights into their structure, performance, and areas for enhancement.

Findings

Most networks are unilaterally structured, affecting knowledge transfer.

Many networks have questionable stability and limited effectiveness.

Few networks effectively share information or utilize network effects.

Abstract

Purpose of our work is to obtain a basic understanding and comparison of the performance and structure of real Knowledge Networks, to identify strengths and weaknesses and to highlight guidelines for improvements. We selected 18 Knowledge Networks from the service sector and 12 networks from the production sector and estimated their Performance and Structure in terms of 19 indices from graph theory. Highlights from our work include: 1) As most networks are unilaterally structured, the direction of knowledge transfer should be taken into account as illustrated in the analysis of clubs and entropy, 2) The stability of most Knowledge Networks is questionable, 3) Few networks are effective in sharing information, while most Knowledge Networks cannot benefit from the network effect, have rather limited capability for coordination, information propagation and synchronization and are not able to integrate Tacit knowledge, 4) Few networks have large cliques which have to be managed with caution as their role may be highly constructive or destructive, 5) While agents with rich connections form clubs, as in most social networks, the poor club effect is not negligible when we take into account the link direction, 6) The directed link analysis of entropy reveals the low complexity-diversification of the Knowledge Networks. In fact the only high entropy network found, has been improved by Knowledge Management Professionals. As most Knowledge Networks underperform, there is plenty of room for further customized analysis in order to improve communication efficiency, coordination, Tacit knowledge dissemination and robustness. This is the first comparative study of real Knowledge Networks in terms of graph theoretic methods.