基于BA网络模型的二部图数据集生成

 1 import random
 2 import networkx as nx
 3 import matplotlib.pyplot as plt
 4 
 5 
 6 def ba_bip_graph(n):
 7     g = nx.DiGraph()
 8     f = open("start.txt")
 9     line = f.readline()
10     repeated_user_nodes = []
11     repeated_item_nodes = []
12     while line:
13         repeated_user_nodes.append(line.strip('
').split(" ")[0])
14         repeated_item_nodes.append(line.strip('
').split(" ")[1])
15         line = f.readline()
16     users = 5
17     user_line = 0
18     item_line = 0
19     while users < n:
20         m1 = random.randint(1, 10)                                   #新增一个user，添加随机0-9个边
21         # print("m1:", m1)
22         item_target = set()
23         while len(item_target) < m1:
24             x = random.choice(repeated_item_nodes)
25             item_target.add(x)
26         label = []
27         f = open("user_name.txt")
28         line = f.readline()
29         while line:
30             label.append(line.strip('
').split(" ")[0])
31             line = f.readline()
32         # print([label[user_line]] * m1)
33         # print(item_target)
34         g.add_edges_from(zip([label[user_line]]*m1, item_target))
35         repeated_item_nodes.extend(item_target)
36         repeated_user_nodes.extend([label[user_line]]*m1)
37         user_line += 1
38         f.close()
39         #新增一个user，新增4个item
40         repeat = 0
41         while repeat < 4:
42             m2 = random.randint(1, 5)
43             # print("m2:", m2)
44             user_target = set()
45             while len(user_target) < m2:
46                 x = random.choice(repeated_user_nodes)
47                 user_target.add(x)
48             label = []
49             f_item = open("item_name.txt")
50             line = f_item.readline()
51             while line:
52                 label.append(line.strip('
').split(" ")[0])
53                 line = f_item.readline()
54             g.add_edges_from(zip([label[item_line]]*m2, user_target))
55             repeated_user_nodes.extend(user_target)
56             repeated_item_nodes.extend([label[item_line]] * m2)
57             item_line += 1
58             f.close()
59             repeat += 1
60         users = users+1
61     return g
62 
63 
64 BA = ba_bip_graph(10)
65 pos = nx.circular_layout(BA)
66 nx.draw(BA, pos, with_labels=True, node_size=20)
67 plt.show()
68 for edge in BA.edges:
69     print(edge)

程序中使用的文件下载地址 https://files-cdn.cnblogs.com/files/xianyuxianyuxian/StartGraph.rar

说明：

在本文中，以start.txt中的数据为起始图，左边添加一个user,右边添加4个item（可以修改）.同时，添加一个点user，随机产生n(0<n<10)条edge,添加一个点item，随机产生n(0<n<5)条edge.函数的形参n是程序结束时user的数目。