AUTHORITIES, HUBS, AND BROKERS IN COMMUNITIES OF PRACTICES

Abstract

Communities of Practice (CoP) have gained importance as a key knowledge management instrument for knowledge sharing in organisations. A CoP is defined as a group of people bound by informal relationships and who share common practices. In this contribution we investigate CoP within a research organisation as knowledge networks where a researcher is considered a node in a knowledge network graph and his/her communication links to others in the CoP as edges. We focus on identifying “important” nodes for knowledge flows such as authorities, hubs and brokers which can be identified by looking at directed graphs. Those CoP members to whom large numbers of people go for advice are authoritative sources. A node with a high in-degree is a good candidate for becoming an authority. A researcher expert with a corresponding high out-degree has a high potential of becoming a hub. Hubs are understood as nodes with high distributive capacity. Knowledge flows are represented as a directed network graph. We calculate the network graphs in this contribution with the software Pajek. For identifying authorities and hubs we apply Kleinberg's Algorithm. For identifying brokers/bridges we apply the measure of betweenness centrality, which measures the shortest path of an expert to all others in the network (developed by White&Borgatti for directed graphs). Experts with a high betweenness centrality have the potential of bridging parts in the network. We link the approach of White&Borgatti with the idea of Kleinberg for detecting researchers with authoritative, hub or broker roles in the CoP knowledge network. Interpreting only the results of the quantitative data without knowing the background of the data could lead to skewed conclusions. Thus the quantitative analysis was complemented with expert interviews in the research organisation. Our results were confirmed and they are helpful but only together with the background information of each node. The introduced methodological approach can be considered as being adequate.