基于马尔科夫模型的移动设备链接预测研究

Research on link prediction based on Markov model for intelligent mobile terminal

受移动设备内存空间和处理器性能的限制,传统的链接预测方法(如传统马尔科夫方法)不再适于移动设备浏览导航.本文以马尔科夫模型为基础,提出基于马尔科夫链和频繁项挖掘相结合的移动设备链接预测模型.将扇形交互界面的思路引入链接预测领域,有效缩小了用户的视觉搜索时间,提高了模型交互效率.移动设备上的实验结果表明,本文提出的预测模型在保证高覆盖率和低复杂度的同时,可以达到较高的预测准确率和预测效率.

Traditional link prediction methods （e.g.,Markov）do not work well on mobile devices be-cause of the limited memory space and poor processor performance of the mobile device.Based on the Markov model,the authors proposed a new link prediction model for mobile devices,which combined the Markov model with the methods of frequent item sets mining.In the paper,the sector display inter-face was introduced into the link prediction field,which effectively reduced the users’visual search time and improved interaction efficiency of the model.In the experiments on mobile devices,the authors proved that the new prediction model obtained high coverage ,low complexity,high prediction accuracy and high prediction efficiency.