基于智能手机用户行为习惯的App使用预测算法研究

PREDICTION ALGORITHM OF APP USAGE BASED ON SMARTPHONE USER BEHAVIOR HABITS

互联网的发展不仅推动了智能手机的普及,而且涌现了大量的手机App。大量被安装在手机中的应用程序不仅增加了用户寻找App的时间和难度,而且占用了手机大量的内存空间,致使手机卡顿,严重影响了用户体验。利用智能手机用户使用App的行为习惯特征,预测用户将要使用的App,并将其应用到智能手机中用以预加载App和智能清理手机内存是解决上述问题的方法之一。在深入了解PrefixSpan算法和Bayesian网络算法的前提下,考虑智能手机用户对每个App的喜爱程度,将其加入到PrefixSpan算法的剪枝步骤中;采用贝叶斯网络算法整合智能手机用户的App使用记录和App使用时长等特征,提出一套新的预测用户下一个将要使用的App的算法——WAPA(Weighted App Prediction Algorithm)。实验证明,该算法预测准确率最高可达86.3%,较其他算法可提高大约4%。

The development of the Internet not only promotes the popularity of smart phone,but also emerges a large of mobile App.A large number of applications installed in the mobile phone increase the time and difficulty for the user to search for App,and occupy a large number of mobile memory space,which leads to the mobile phone lagging.This phenomenon has seriously influenced the user experience.It is one of the ways to solve the above problem by using the behavior and habit characteristics of App for smart phone users to predict the App that the user will use,and apply it to the smart phone to preload the App and optimize the memory of the mobile phone.Under the premise of understanding the PrefixSpan algorithm and the Bayesian network algorithm.We considered the preference of smartphone users for each App,and added it to the pruning step of the PrefixSpan algorithm.Then,we integrated user s App usage record and the time for using App with Bayesian network algorithm.Therefore,a new set of App algorithm WAPA(Weighted-App-Prediction-Algorithm) was proposed.Experiments show that the algorithm has a prediction accuracy of up to 86.3%,which is about 4% higher than other algorithms.