Abstract:

Dynamic network link prediction is an important topic in the field of network data mining. It predicts the future network structure status according to the previous network structure. At present, static link prediction has been fully studied, but the research of dynamic link prediction is little. According to the characteristics of network links, this paper introduces the matrix completion method into the dynamic link prediction problem. Further inspired by the kernel matrix decomposition, this paper establishes the kernel matrix completion model. Mapping data into high-dimensional spaces makes nonlinear relationships into linear relationships, which makes the model can handle more complex network structures. The effectiveness of applying the matrix completion method and kernel method to dynamic link prediction is verified by experiments on three public network datasets.

Key words: dynamic link prediction, matrix completion, matrix decomposition, kernel method

摘要：

动态链接预测是网络数据挖掘领域的一个重要课题，主要原理是根据以往的网络结构预测未来的网络结构状态。目前，静态链接预测已得到充分研究，但对动态链接预测的研究却比较稀少。根据网络链接的结构特点，将矩阵补全方法引入动态链接预测问题中，进一步受核矩阵分解的启发，建立了核矩阵补全模型，将数据映射到高维空间中，使得链接中的非线性关系转化为线性关系，从而使得模型能够处理更复杂的网络结构。通过在三个公开网络数据集上进行实验，验证了矩阵补全优化方法和核方法在动态链接预测中的有效性和准确性。

关键词: 动态链接预测, 矩阵补全, 矩阵分解, 核方法