pagerank 算法入门

http://blog.csdn.net/midgard/article/details/7061721

这篇文章很不错。

#include <vector>
#include <set>
#include <string>
#include <iostream>

using namespace std;

// use graph store webpage, weight representlink times
class Node {
public:
    explicit Node(string name, double pr = 1) :
            name_(name), page_rank_(pr) {
    }

    ~Node() {
        linkin_nodes_.clear();
    }
    void InsertLinkdInNode(Node* node) {
        //如果没有链接
        if (linkin_nodes_.find(node) == linkin_nodes_.end()) {
            linkin_nodes_.insert(node);
        }
        node->InsertLinkOutNode(this);
    }

    void InsertLinkOutNode(Node* node) {
        //如果没有链接
        if (linkout_nodes_.find(node) == linkout_nodes_.end()) {
            linkout_nodes_.insert(node);
        }
    }

    void InsertLinkdInNode1(Node* node) {
        //如果没有链接
        if (linkin_nodes_.find(node) == linkin_nodes_.end()) {
            linkin_nodes_.insert(node);
        }

    }

    void InsertLinkOutNode1(Node* node) {
        //如果没有链接
        if (linkout_nodes_.find(node) == linkout_nodes_.end()) {
            linkout_nodes_.insert(node);
        }
        node->InsertLinkdInNode1(this);
    }

    double GetPageRank() {
        return page_rank_;
    }

    void SetPageRank(double pr) {
        page_rank_ = pr;
    }

    double CalcRank() {
        double pr = 0;
        set<Node*>::const_iterator citr = linkin_nodes_.begin();
        for (; citr != linkin_nodes_.end(); ++citr) {
            Node * node = *citr;
            pr += node->GetPageRank() / node->GetOutBoundNum();
        }
        return pr;
    }

    size_t GetOutBoundNum() {
        return linkout_nodes_.size();
    }

    size_t GetInBoundNum() {
        return linkin_nodes_.size();
    }

    void PrintNode() {
        cout << "Node:" << name_ << " 's pagerank is: " << page_rank_ << endl;
    }
private:
    string name_;
    set<Node*> linkin_nodes_;
    set<Node*> linkout_nodes_;
    double page_rank_;
};

class PageRank {
public:
    PageRank(double q = 0.85);
    ~PageRank(void);
    void Calc(vector<Node*> & nodes, int n);
    double Calc(Node* node);
    void PrintPageRank(vector<Node*> & nodes);
private:
    double q_; //阻尼系数
};

#include "PageRank.h"
#include <iostream>

PageRank::PageRank(double q) :
        q_(q) {
    // q_ must < 1
}

PageRank::~PageRank(void) {
}

// 迭代计算n次
void PageRank::Calc(vector<Node*> & nodes, int n) {
    for (int i = 0; i < n; ++i) {
        vector<Node*>::const_iterator citr = nodes.begin();
        for (; citr != nodes.end(); ++citr) {
            Node * node = *citr;
            Calc(node);
        }
    }
}

void PageRank::PrintPageRank(vector<Node*> & nodes) {
    double total_pr = 0;
    vector<Node*>::const_iterator citr = nodes.begin();
    for (; citr != nodes.end(); ++citr) {
        Node * node = *citr;
        node->PrintNode();
        total_pr += node->GetPageRank();
    }
    cout << "Total PR:" << total_pr << endl;
}

double PageRank::Calc(Node * node) {
    double pr = node->CalcRank();
    if (pr < 0.00000000000000000000001 && pr > -0.00000000000000000000001) //pr == 0
            {
        pr = 1 - q_;
    } else {
        pr = pr * q_ + 1 - q_;
    }
    node->SetPageRank(pr);
    return pr;
}

#include <iostream>
#include <vector>
#include <string>
#include <map>
#include "PageRank.h"

using namespace std;

void InitGraph(vector<Node*> & nodes) { // 邻接表存储方式
                                        // example 1
    Node * a = new Node("A");
    Node * b = new Node("B");
    Node * c = new Node("C");
    Node * d = new Node("D");
    nodes.push_back(a);
    nodes.push_back(b);
    nodes.push_back(c);
    nodes.push_back(d);
    // link in node
    // a <- b, c, d
    a->InsertLinkdInNode(b);
    a->InsertLinkdInNode(c);
    a->InsertLinkdInNode(d);
    // b <- d
    d->InsertLinkOutNode1(b);
//b->InsertLinkdInNode(d);
    // c <- b, d
    c->InsertLinkdInNode(b);
    c->InsertLinkdInNode(d);
}

void TestPageRank() {
    // build graph
    vector<Node*> nodes;
    InitGraph(nodes);
    PageRank pr;
    // 迭代计算5次 pagerank
    pr.Calc(nodes, 80);
    pr.PrintPageRank(nodes);
}

int main(int argc, const char ** argv) {
    TestPageRank();
    return 0;
}