G面经prepare: Friends Recommendation - neverlandly - 博客园

G面经prepare: Friends Recommendation - neverlandly - 博客园

想想如果你用linkedin或者facebook， 给你一个人和他的朋友关系网，你会怎么给一个人推荐朋友

一个例子就是A-B, A-C, B - D, B - E, C - D，这个时候问我应该推荐谁给A，我说D,因为他是BC的共同好友，而E只是B的好友,到这我才明白干啥，就是给一个图和里面的一个节点A,用bfs从A出发，找出第二层中indegree度数最大节点
用HashMap<Character, HashSet<Character>>来建图
用HashMap<Character, Integer> SndIndegree来表示第二层indegree数目
用maxIndegree记录第二层Indegree最大值
用res记录第二层Indegree最大者

 5     public char recommend(char tar, char[][] arr) {
 6         HashMap<Character, HashSet<Character>> graph = new HashMap<Character, HashSet<Character>>();
 7         HashMap<Character, Integer> SndIndegree = new HashMap<Character, Integer>();
 8         
 9         
10         //build graph
11         for (char[] edge : arr) {
12             if (!graph.containsKey(edge[0])) graph.put(edge[0], new HashSet<Character>());
13             if (!graph.containsKey(edge[1])) graph.put(edge[1], new HashSet<Character>());
14             graph.get(edge[0]).add(edge[1]);
15             if (!SndIndegree.containsKey(edge[0])) SndIndegree.put(edge[0], 0);
16             if (!SndIndegree.containsKey(edge[1])) SndIndegree.put(edge[1], 0);
17         }
18         
19         Queue<Character> queue = new LinkedList<Character>();
20         HashSet<Character> visited = new HashSet<Character>();
21         int level = 0;
22         queue.offer(tar);
23         visited.add(tar);
24         int PNum = 1;
25         int CNum = 0;
26         int maxIndegree = 0;
27         char res = '\0';
28         while (!queue.isEmpty()) {
29             char cur = queue.poll();
30             PNum--;
31             for (Character neigh : graph.get(cur)) {
32                 if (level+1 == 2) {
33                     if (neigh == tar) continue;
34                     int curIndegree = SndIndegree.get(neigh)+1;
35                     if (curIndegree > maxIndegree) res = neigh.charValue();
36                     SndIndegree.put(neigh, curIndegree);
37                 }
38                 else { //not second level
39                     if (!visited.contains(neigh)) {
40                         queue.offer(neigh);
41                         CNum++;
42                         visited.add(neigh);
43                     }
44                 }
45             }
46             if (PNum == 0) {
47                 PNum = CNum;
48                 CNum = 0;
49                 level++;
50             }
51         }
52         return res;
53     }
59     public static void main(String[] args) {
60         // TODO Auto-generated method stub
61         Solution sol = new Solution();
62         char res = sol.recommend('A', new char[][]{{'A','B'},{'A','C'},{'B','D'},{'B','E'},{'C','D'},{'B','A'},{'C','A'}});
63         System.out.println(res);
64     }

Read full article from G面经prepare: Friends Recommendation - neverlandly - 博客园