We talked about the communities this week. Community is a set of actors interacting with each other. If a there is a set of people without interaction we cannot say they form a community.
There are four community detection methods; Node-centric community, group-centric community, network-centric community, hierarchy-centric community.
Each node in a group satisfies certain properties. We make node-centric community detection based on these properties. This method is commonly used in traditional social network analysis. In this approach nodes satisfy different properties. One is complete mutuality. A clique is a maximal complete subgraph of three or more nodes all of which are adjacent to each other. Reachability of members is another issue considering nodes. Any node in a group should be reachable in k hops. A k-clique is a maximal subgraph in which the largest geodesic distance between any nodes is smaller than or equal to k. A k-club graph is a substructure of diameter smaller than or equal to k. In node-centric community each node should have a certain number of connections to nodes within the group. In a k-core graph each node connects to at least k members within the group.
To be able to make group-centric community detection we consider the connections within a group as whole.
To form a group we need to consider the connections of the nodes globally in network-centric community detection. We partition the network into disjoint sets. While partitioning we consider some features like node similarity,latent space model, block model approximation, cut minimization and modularity maximization. Node similarity is defined by how similar their interaction patterns are. Two nodes are structurally equivalent if they connect to the same set of actors. For huge networks we apply k-node clustering algorithms to seperate the network.
Finally the goal in hierarchy-centric community detection is to build a hierarchical structure of communities based on network topology.
No comments:
Post a Comment