2014-11-20:PageRank学习记录

Stanford CS224W Problem Set 0

昨天做的是第二题，给出一个数据集（stackoverflow-Java.txt），运用相关工具求出

• 弱连通分量（weakly connected components）
• 最大弱连通分量中节点个数、边的个数
• PageRank算法排名最高的前三个的ID以及score
• HITS算法排名前三个hubs以及authorities

题目比较简单，主要考查的是对工具的掌握情况，具体讲解在以下代码中说明。

##################################################################
# Project: Finding experts on the java programming
#           language on Stack-Overflow 
# Tool: SNAP 
# Version:1.0
# Date: 2014-11-19
# Author: Chuanting.Zhang
# Email: chuanting.zhang@gmail.com 
# Ps: Written by Burning at SDU
###################################################################
import snap;
from snap import PNGraph;
from audioop import reverse

#加载进数据，数据包含N行两列
Graph = snap.LoadEdgeListStr(PNGraph, 'stackoverflow-Java.txt', 0, 1)
print 'Number of nodes in the network: %d' % Graph.GetNodes()

#涉及到图的分量时要先定义一个变量来存储连通分量
Components = snap.TCnComV()
snap.GetWccs(Graph, Components)
print 'Number of weakly connected components in the network: %d' %Components.Len()

#最大弱联通分量
mxWcc = snap.GetMxWcc(Graph)# Return a graph 
print 'Number of edges and nodes in the largest weakly connected components: %d %d' %(mxWcc.GetEdges(), mxWcc.GetNodes())

#PageRank算法
#PageRank返回的是节点ID以及排名值，所以把它存储到一个哈希表中
pRankH = snap.TIntFltH()
snap.GetPageRank(Graph, pRankH)

#逆序排序
slist = sorted(pRankH, key = lambda key: pRankH[key], reverse = True)
for item in slist[:10]:
    print 'id: %7s, pagerank: %.6f' %(item,pRankH[item])

NIdHubH = snap.TIntFltH()
NIdAuthH = snap.TIntFltH()

#HITS算法
snap.GetHits(Graph, NIdHubH, NIdAuthH)
slistHub = sorted(NIdHubH, key = lambda key: NIdHubH[key], reverse = True)
for item in slistHub[:3]:
    print 'id: %7s, hub: %.6f' %(item, NIdHubH[item])
    
slistAurth = sorted(NIdAuthH, key = lambda key: NIdAuthH[key], reverse = True)
for item in slistAurth[:3]:
    print 'id: %7s, authority: %.6f' %(item, NIdAuthH[item])

 题目虽然简单，但是里面涉及的算法还是需要搞明白。