An application prediction method for context awareness and adaptation in mobile systems

Abstract

The invention discloses an application prediction method used for context awareness and adaptation in a mobile system. The application prediction method comprises the following steps that step 1, record information effective for improving the mobile application prediction accuracy is extracted from a large amount of data collected by mobile equipment to act as a training set; step 2, the training set extracted based on the step 1 is trained by using an unbalanced Bayesian model, and an application use probability model is obtained through training; and step 3, the application to be used is predicted by applying the use probability model based on the awared current context information, the predicted application to be used and the actually used application are compared to obtain the current prediction accuracy, and then the size of each prediction period is adaptively adjusted according to the current prediction accuracy by using a flexible algorithm so as to reduce the training cost. The technical effects of the application prediction method are that the application prediction accuracy can be enhanced and the training cost of the prediction model can be effectively reduced.

Description

technical field [0001] The invention belongs to the technical field of mobile system application prediction, and in particular relates to a context-aware and self-adaptive application prediction. Background technique [0002] Mobile applications on mobile systems are becoming more and more numerous and larger, increasing the supply pressure on power and memory. Application prediction refers to the prediction of upcoming applications, which can improve user experience. However, there are two problems in the existing application prediction methods: one is low prediction accuracy; the other is high training cost. Especially in some fields of application prediction, such as memory management and mobile application pre-launch, which require high application prediction accuracy and low training costs, existing application prediction methods cannot meet the requirements. [0003] The application prediction method based on the Bayesian model can improve the prediction accuracy. Th...

AI Technical Summary

Problems solved by technology

However, there are two problems in the existing application prediction methods: one is the low prediction accuracy; the other is the high training cost

figure 2 The experimental results of 5 users are shown. The arrow in the figure points to the vertical axis used by the curve, and the abscissa indicates the size of the training set (the unit is the number of records in one week). The application prediction accuracy will first change with the training set. large and rises, and subsequently, the applied prediction accuracy drops

In other words, using a large training set does not yield high prediction accuracy

Images