Networkx中无向图连通分量节点度中心性均小于1的原因问询

Hey there! Let's unpack what's happening with your degree centrality scores and your betweenness centrality implementation.

Why All Degree Centrality Values Are < 1 #

First, let's clarify how NetworkX calculates normalized degree centrality (the default behavior of nx.degree_centrality()):

• The formula used is node_degree / (total_nodes - 1), where total_nodes is the number of nodes in your connected component GC.
• A value of 1 would only occur if a single node is connected to every other node in the component (a perfect star graph). Since all your values are below 1, this just means no node in your component is fully connected to every other node—this is extremely common in real-world networks (think social networks, where no one knows every single person in a large group).
• Your highest score of ~0.693 tells us the most connected node is linked to roughly 69% of the other nodes in the component. For example, if your component has 100 nodes, this node would have a degree of ~69.

Checking Your Betweenness Centrality Code #

Your code for calculating and sorting betweenness centrality is valid, but here are a few notes to make it more informative:

• The default nx.betweenness_centrality() also returns normalized scores (divided by (n-1)(n-2)/2 for undirected graphs). If you want raw, unnormalized counts of shortest paths passing through each node, add the normalized=False parameter:

  bcGC_dict = dict(nx.betweenness_centrality(GC, normalized=False))
  

• To see both the node IDs and their corresponding betweenness scores in your top 11 list, modify your sorting step to include the values:

  ordered_bcGC = sorted(bcGC_dict.items(), key=lambda x: x[1], reverse=True)
  print("\nTop 11 betweenness centrality nodes (connected component):")
  for node, score in ordered_bcGC[:11]:
      print(f"Node: {node}, Betweenness Score: {score:.4f}")
  

• Keep in mind that a node with high degree centrality doesn't always have high betweenness centrality. Some nodes act as "bridges" between different parts of the graph—they might have low degree but sit on most of the shortest paths between distant nodes, leading to a high betweenness score.

Common Follow-Up Scenarios (Based on Your Partial Query) #

Since you mentioned "是否有可能..." (is it possible that...), here are a few common scenarios to consider:

• Is it possible for a node with low degree to have high betweenness? Yes—this is typical of bridge nodes that connect disjoint subgraphs within your component.
• Is there a mistake if top degree and betweenness nodes don't overlap? No—this is a normal property of many real-world graphs, where different nodes serve distinct structural roles.

内容的提问来源于stack exchange，提问作者MegaM