¶ Description
Compute significance level of each arc.
¶ Parameters
Parameters:
- Column with id A Column representing the source node ID.
- Column with id B Column representing the destination node ID.
- Weight of connection A<->B Column containing the connection weight (e.g., number of interactions).
- Maximum value for a node id Optional upper limit for node ID values (used for memory-optimized matrix).
¶ About
The SignificanceLevel action computes the statistical significance level of each weighted connection between nodes in a network graph, based on node identifiers and connection weights. This is especially useful in network analysis tasks, where it is important to identify the relevance or strength of individual edges based on their weight relative to the overall graph structure.
¶ Use Case
Use this action in scenarios where you analyze large-scale networks (e.g., infection spread, social interaction, or transportation graphs) and want to understand which links (edges) between nodes are most significant. The output adds a new column significanceLevel, which indicates the normalized significance of the connection between two nodes.
¶ How It Works
This action reads a table containing pairwise connections between entities (nodes), computes the total strength of outgoing and incoming links for each node, and evaluates how significant each individual link is based on its weight and the global node ID distribution.
The result is a new significanceLevel column with a value between 0 and 1:
- 0 means the connection is statistically insignificant.
- 1 indicates maximum relative importance within the network.
¶ Input Requirements
The input table must contain:
- A pairwise connection between two node IDs (columns A and B).
- A numeric weight for the connection.
Example input:
| A | B | Weight |
|---|---|---|
| 1 | 5 | 0.8 |
| 1 | 2 | 0.9 |
| 2 | 3 | 0.6 |
¶ Output
The output will include all input columns plus an additional significanceLevel column:
¶ Notes
- Ensure node IDs are numeric and within the range set in
Maximum value for a node id. - If the max value is set too low, errors may occur or results may be truncated.
- The output can be interpreted as a normalized significance score per edge, useful for filtering or ranking connections in the graph.
¶ Where This Can Be Used
- Epidemiology: Identify significant infection pathways in disease-spread networks.
- Social Networks: Detect meaningful connections in a large graph of users or communities.
- Supply Chains: Highlight critical shipment or transit routes.
- Knowledge Graphs: Assess the strength of entity relationships.
