Knowledge communities feature specific structural properties. These high-level properties are important for two reasons:
1. they are oftentimes stable characteristics specific to each system that inform its organization.
2. they emerge from the local interactions between actors and content, yet they may be structuring for future micro-dynamics.
At the micro level, dynamics exhibit a great number of regularities. Besides classical effects regarding social structure (rich get richer effect), we observe, for example, that there are strong relations between the likeliness that two agents are to get connected and the contents they share. We observe various patterns according to the kind of system under study (see Social and Semantic coevolution in knowledge networks).
But not all knowledge networks are based on dyadic interaction between pairs of agents. Collaboration networks, for example, are basically built from the aggregation of grouping of co-authors ; these groups can gather more than 2 individuals. The real collaboration dynamics should then be appraised through a hypergraphic approach which enables to introduce meso-level characteristics when looking for regularities in the team formation process.
Empirical investigation on blog network also challenge the usual metaphor of small world networks. From an individual point of view, online interactions are overwhelmingly produced in a local arena which seems to favor balkanization process.