Identifying Communities of Practice through Ontology Network Analysis

Increasingly, organizations are harnessing communities of practice to carry out important knowledge management functions.3 However, a significant first step is identifying the community, which often doesn’t designate itself as such, and its members, who don’t know they belong! So, this step involves determining which people in a community of practice have common interests in particular practices or functions and producing sets or clusters of related individuals. Community identification traditionally demands heavy resources and often includes extensive interviewing. In this article, we describe Ontocopi (OntologyBased Community of Practice Identifier), a tool to help identify communities. Ontocopi lets you infer the informal relations that define a community of practice from the presence of more formal relations. For instance, if A and B have no formal relation but they have both authored papers with C (formal relation), they might share interests (informal relation). Because Ontocopi works in this way, we cannot claim without qualification that it identifies communities of practice. Significant informal relations might have little or no connection to the formal ones. Here, we refer to the networks uncovered by Ontocopi as COPs and to informal social networks as communities of practice. We work under the assumption that COPs are sometimes decent proxies for communities of practice. Ontocopi We developed Ontocopi within the Advanced Knowledge Technologies (AKT) project4 for ontologybased network analysis (ONA), which finds sets of instances associated with a selected instance in a knowledge base. (See the related sidebar.) By ontology, we mean combining a taxonomic structure of classes and relations with the knowledge base that results from instantiating the classes with domain objects. If you suppose that such an ontology represents a domain’s objects and relations, you can analyze the connections between the objects. Ontocopi uses ontological relations to discover connections between objects that the ontology only implicitly represents. For example, the tool can discover that two people have similar patterns of interaction, work with similar people, go to the same conferences, and subscribe to the same journals. Using an ontology to analyze such networks provides you with semantics for classes and relations. So, during analysis, you can select targeted relations for the community of practice and increase their weight in the algorithm and assign low or zero weights to unimportant relations. However, choosing the ontology is an important step because its content determines ONA’s effectiveness. For example, the papers people publish are likely to be important for determining their interests, but if their publications This article describes