| Smart phone application operators mainly provide three kinds of applications to mobile users currently: Native App,Web App and Hybrid App.The Native App is a mobile application early developed as mobile application software.Usually,mobile users almost need not to spend a lot of time to install,update Web App.It turns out that Web App is more convenient.However,a large amount of data indicate that Web App has not been widely used compared with Native App and Hybrid App.This phenomenon has attracted the attention of many researchers.It has been found that Web App has some problems such as low performance of message transmission compared with Native App and Hybrid App.When users access Web App using smart phone,large volume of data is transferred including many duplicated data,leading to long waiting time and repeating operation problems.To resolve the problem of low performance in Web App usage,we treat Web App pages as a complex graph.Redirecting between pages is similar as relationship between nodes which is also called edges of the graph.We realize the prototype system of pre-fetching mechanism based on user behavior.To get rid of the interrupting of other factors,we build a Web App example and then get the web logs to mine the user behavior pattern.We build relationship matrix of the graph and assign calculated user behavior value to it.The model will supply a prediction of user behavior and pre-fetch it to local storage by evaluating possible behavior.Loading data from local storage will improve Web App performance since we could avoid sending request to remote server and waiting the response.In the phase of the performance experiment,we use WebPageTest tool to test Web App performance firstly and explore real-time performance parameters to the detail using ADB tool.Finally,we discuss how to improve Web App performance and efficiency using our method and conclude the work routine to improve Web App performance according to our model and the test results. |
PDF Full Text Request