The closeness functions return closeness and k-degree scores for vertexes in a graph.
Closeness and k-degree scores are fundamental distance-based centrality metrics used in network structure analysis. Both measure the time needed to spread information from a source vertex to a set of target vertices.
The closeness score calculated by the ApproximateCloseness function is defined for each vertex v as the inverse of the approximate sum of the shortest distances from v to all other reachable vertices u.
The k-degree score is defined for vertex v as the number of vertices whose distance from v is less than or equal to k.
| Function | Description |
|---|---|
| Closeness (ML Engine) | Returns closeness and k-degree scores for each specified source vertex in graph. Graph can be directed or undirected, weighted or unweighted. Applies classical definition of closeness to connected graphs and alternative definition to disconnected graphs. For large graph, you can apply function to random sample of specified target vertices to get efficient approximation of closeness and k-degree scores. |
| ApproximateCloseness (ML Engine) | Highly scalable, approximate (near-accurate) algorithm for finding closeness centrality and k-degree scores for all vertices in unweighted graph. Uses fast, memory-efficient cardinality estimation measure called HyperLogLog to achieve superior performance. Teradata recommends this function for large-scale graphs with up to 2 billion vertices. |