HITS算法分析与实现

HITS 算法分析与实现

本文转载自：https://blog.csdn.net/hguisu/article/details/8013489

1、内容详细介绍参见以下博文

https://blog.csdn.net/hguisu/article/details/8013489

2、代码实现

from pygraph.classes.digraph import digraph
from math import sqrt


class HITSIterator:
    __doc__ = '''计算图中的Hub，Authority值'''

    def __init__(self, dg):
        self.max_iterators = 100  # 最大迭代次数
        self.min_delta = 0.0001  # 判断是否迭代结束
        self.graph = dg

        self.hub = {}

        self.authority = {}

        for node in self.graph.nodes():
            self.hub[node] = 1
            self.authority[node] = 1

    def hits(self):
        """
        计算每个页面的Hub,Authority值
        """
        if not self.graph:
            return

        flag = False

        for i in range(self.max_iterators):
            change = 0.0  # 变化量
            norm = 0
            tmp = {}
            tmp = self.authority.copy()

            # 计算每个页面的authority值
            for node in self.graph.nodes():
                self.authority[node] = 0
                for incident_page in self.graph.incidents(node):  # 入射边
                    self.authority[node] += self.hub[incident_page]
                norm += pow(self.authority[node], 2)

            # 标准化
            norm = sqrt(norm)
            for node in self.graph.nodes():
                self.authority[node] /= norm
                change += abs(tmp[node] - self.authority[node])