APP PERMISSION CLASSIFICATION : STATIC AND DYNAMIC METHODS

Abstract

It is no secret that Android is one of the most widely used smartphone operating systems globally, boasting a staggering 2.5 billion active users. However, data secu¬rity has become a crucial aspect of smartphone usage with the increasing reliance on smartphones to store sensitive personal information. Unfortunately, many apps tend to collect user data without the user's knowledge or consent, which can harm data security. To counter this, Android has included an inbuilt security feature called app permission to enable users to control app access. This feature enables users to grant or decline app access to various phone features such as camera, microphone, and location data. The study proposes two methodology for classification app permissions. Firstly, static novel three-tiered system called APEC (App Permission Classification with Efficient Clustering).The study provides an insightful analysis of app permissions using a dataset of 2 million app permissions and app categories from the Google Play Store. The static novel three-tiered system called APEC (App Permission Classification with Efficient Clustering) aims to determine the safety of app permis¬sions based on their usage frequency within specific app categories. This system categorizes apps into three levels, clustering, approval, and classification, to ensure users can select appropriate permissions and developers can establish the minimum requirements for their apps to function smoothly. To accomplish this, APEC uses DBSCAN clustering to group apps based on their respective categories and evalu¬ate the safety of their permissions. Furthermore, it employs the Decision Tree and Random Forest machine learning algorithms to classify new app permissions as safe or unsafe. The proposed system achieved an impressive accuracy rating of 93.8% and 95.8% using the Decision Tree and Random Forest algorithms, respectively. Secondly, a dynamic methodology App Permission Classification Dynamic Model (APCM) based on APEC with added features and capabilities. It keeps track of the app permissions requested by various apps using a dataset of 2 million apps from the Google Play Store to train a random forest-based model using DBSCAN clustering. The APCM analyses each permission request's frequency in a category and creates a frequency map accordingly. This model is critical in rating the app permission as safe or unsafe by using DBSCAN clustering and predicting using a random forest machine learning algorithm. The APCM further enhances the results by using PSO¬BO optimization over the random forest, which is considered a highly accurate approach. The APCM provides a better experience and ensures that users can use apps safely without any concerns about their security. By analyzing the frequency of each permission request in a category, the APCM can identify safe and unsafe permissions and ensure that users can use apps without any worries. This model considers user preferences while classifying apps using the Dataset, ensuring a more personalized and satisfying user experience. The APCM has achieved an impressive accuracy of 87% in classifying an app's permission as safe or unsafe. Overall, the APCM is a highly innovative and effective approach that can help users ensure their apps are safe and secure. The use of DBSCAN clustering and random forest machine learning algorithms, and PSO-BO optimization ensure that the APCM is highly accurate and can provide a reliable classification of app permissions. This model is also highly user-friendly, ensuring users can use apps safely and securely without concerns about their security or data privacy.