Clustering and community detection algorithms in Neptune Analytics
Clustering algorithms evaluate how nodes are clustered in communities, in closelyknit sets, or in highly or loosely interconnected groups.
These algorithms can identify meaningful groups or clusters of nodes in a network, revealing hidden patterns and structures that can provide insights into the organization and dynamics of complex systems. This is valuable in social network analysis and in biology, for identifying functional modules in proteinprotein interaction networks, and more generally for understanding information flow and influence propagation in many different domains.
Neptune Analytics supports these community detection algorithms:

wcc – The Weakly Connected Components (WCC) algorithm finds weaklyconnected components in a directed graph. A weaklyconnected component is a group of nodes where every node in the group is reachable from every other node in the group if edge direction is ignored.
Identifying weaklyconected components helps in understanding the overall connectivity and structure of the graph. Weaklyconnected components can be used in transportation networks to identify disconnected regions that may require improved connectivity, and in social networks to find isolated groups of users with limited interactions, and in webpage analysis to pinpoint sections with low accessibility.

wcc.mutate – This algorithm stores the calculated component value of each given node as a property of the node.

labelPropagation – Label Propagation Algorithm (LPA) is an algorithm for community detection that is also used in semisupervised machine learning for data classification.

labelPropagation.mutate – Label Propagation Algorithm (LPA) is an algorithm tha assigns labels to nodes based on the consensus of their neighboring nodes, making it useful for identifying groups. Label propagation can be applied in social networks to find groups, and in identity management to identify households, and in recommendation systems to group similar products for personalized suggestions. It can also be used in semisupervised machine learning for data classification.

scc – The Strongly Connected Components (SCC) algorithm identifies maximally connected subgraphs of a directed graph, where every node is reachable from every other node. This can provide insights into the tightly interconnected portions of a graph and highlight key structures within it. Strongly connected components are valuable in computer programming for detecting loops or cycles in code, in social networks to find tightly connected groups of users who interact frequently, and in web crawling to identify clusters of interlinked pages for efficient indexing.

scc.mutate – This algorithm finds the maximally connected subgraphs of a directed graph and writes their component IDs as a new property of each subgraph node.