# -*- coding: utf-8 -*-
"""网络结构评估指标。
提供网络结构特征的评估指标函数。
"""
from typing import List, Set, Dict, Optional, TYPE_CHECKING
import numpy as np
from collections import Counter
if TYPE_CHECKING:
from ..graph import IMGraph
[docs]
def distribution_entropy(
graph: 'IMGraph',
seeds: Set[int]
) -> float:
"""计算种子节点在度分布上的熵。
熵值越高表示种子节点分布越均匀。
Args:
graph: 图对象。
seeds: 种子节点集合。
Returns:
float: 分布熵，范围 [0, 1]。
- 1.0 表示分布完全均匀
- 0.0 表示分布完全集中
Example:
>>> from pynetim.evaluation import distribution_entropy
>>> entropy = distribution_entropy(graph, seeds)
>>> print(f"Distribution entropy: {entropy:.4f}")
"""
if len(seeds) == 0:
return 0.0
degrees = graph.batch_out_degree(list(seeds))
degree_counts = Counter(degrees)
total = len(seeds)
probabilities = [count / total for count in degree_counts.values()]
entropy = -sum(p * np.log(p) for p in probabilities if p > 0)
max_entropy = np.log(len(degree_counts)) if len(degree_counts) > 1 else 1.0
return entropy / max_entropy if max_entropy > 0 else 0.0
[docs]
def local_clustering(
graph: 'IMGraph',
seeds: Set[int]
) -> float:
"""计算种子节点的局部聚类系数。
评估种子节点周围网络的紧密程度。
Args:
graph: 图对象。
seeds: 种子节点集合。
Returns:
float: 平均局部聚类系数，范围 [0, 1]。
Example:
>>> from pynetim.evaluation import local_clustering
>>> clustering = local_clustering(graph, seeds)
>>> print(f"Local clustering: {clustering:.4f}")
"""
from .seed_quality_metrics import clustering_coefficient
return clustering_coefficient(graph, seeds)
[docs]
def reachability(
graph: 'IMGraph',
seeds: Set[int]
) -> float:
"""计算种子节点的可达性。
可达性 = 从种子节点可达的节点数 / 网络总节点数
Args:
graph: 图对象。
seeds: 种子节点集合。
Returns:
float: 可达性，范围 [0, 1]。
Example:
>>> from pynetim.evaluation import reachability
>>> reach = reachability(graph, seeds)
>>> print(f"Reachability: {reach:.2%}")
"""
if len(seeds) == 0:
return 0.0
reachable_nodes = set(seeds)
for seed in seeds:
visited = {seed}
queue = [seed]
while queue:
node = queue.pop(0)
for neighbor in graph.out_neighbors(node):
if neighbor not in visited:
visited.add(neighbor)
queue.append(neighbor)
reachable_nodes.update(visited)
return len(reachable_nodes) / graph.num_nodes