基于Hadoop的跨社交网络局部时序链路预测算法

Hadoop-based local timing link prediction algorithm across social networks

为了提高跨社交网络局部时序链路预测的精度和平稳性,提出了基于Hadoop的跨社交网络局部时序链路预测算法。该算法选取了6种跨社交网络节点相似性指标,采用Hadoop的核心组件MapReduce设计了一种并行运算模型,分割处理跨社交网络内的海量并行数据,降低了运算复杂度。运用基于MapReduce并行运算模型的局部时序链路预测算法和所选取的节点相似性指标,获得网络内点对间的预测分数值,实现了跨社交网络局部时序链路预测。实验结果表明,本文算法的预测精度较高,且能够保持平稳的预测状态,具有较好的综合预测性能。

In order to improve the accuracy and stability of local time series link prediction for cross social network,a Hadoop based local time series link prediction algorithm is proposed. This algorithm selects six cross social network node similarity indicators,and designs a parallel computing model using the core component MapReduce of Hadoop,which can segment and process the massive parallel data in cross social network,and reduce the computational complexity. Based on this,the local time series link prediction algorithm based on MapReduce parallel operation model is used to obtain the point-to-point prediction score in the network by using the selected node similarity index,so as to realize the prediction of cross social network local temporal link. The experimental results show that the prediction accuracy of the proposed algorithm is high, it can maintain a stable prediction state, and has good comprehensive prediction performance.