Design And Implementation Of A Mobile Applications Classifier

At present,the amounts of domestic mobile applications markets more than 200,but there are no strict standard to manage mobile applications.Mobile applications classification research is based on the background,which is important content of market segmentation.Mobile applications classification is significant not only for users and phone manufacturers but also for the relevant regulatory authorities.Intuitive,mobile applications classification research can help users understand their preferences;for mobile phone manufacturers can tap out the market potential of mobile applications around the effective market delivery;regulators can understand the development of mobile applications markets through the classified data.This thesis studies the deep relationship between the Chinese mobile applications and the code information of applications,including the following contents:1.Aiming at the problem of synonyms in mobile applications metadata,analyzing the semantic meaning of words,identifying the semantic related terms and eliminating the influence of Chinese synonyms on classification,the average accuracy rate in support vector machine experiments increased by 4.92%.The experimental results show that the analysis of metadata synonyms can improve the recognition rate of mobile applications.2.According to the problem of vocabulary association in mobile applications description information,the deep learning network structure is analyzed to understand the context.The average accuracy of classification is 82.42%,which is 7.73%higher than that of Naive Bayes.It proves that the deep learning network trained in this thesis is more suitable for analysis mobile applications metadata.3.In order to solve the problem of code information in mobile applications,this thesis analyzes the applications binary file of mobile applications,extracts the information of the mobile applications,classifies the mobile applications according to the permission,compares the classification of permission and the result of random classification,and indicates that the permission information can be well used in mobile applications classification.