Latent Relation Analysis based Discovering Fraudulent Ranking Identification on Mobile Web Apps

Abstract

Objective: The main objective of this work is finding a fraudulent behaviour of mobile apps where mobile app developers may generate fraudulent evidences for providing an top ranking for them. The primary goal of this work is to find out the fraudulent evidences present in the ranked mobile apps. And also this work aims to filter the mobile fraudulent ranking behaviour based on the semantic relation present among the evidences of mobile apps. Method: In the existing work, leading session methodology is introduced to leverage the fraudulent ranking activities. And also the three types of evidences are analysed and aggregated to detect the fraudulent behaviour. However this method cannot consider the relationship among the evidences that are analysed for malicious behaviour detection which will lead to an inefficient detection of fraudulent ranking behaviour. To overcome this problem, in this work latent semantic relationship among the evidences are analysed and based on the relationship exists among them fraudulent ranking behaviour is detected. This is done by constructing the vector of entities which can capture the degree of association present among the concept vectors. Application/Improvements: This proposed research methodology would be more helpful in the mobile app markets where the number of apps developed for the specific purpose has been increased considerably. In this situation, it is required to provide truthful and most popular mobile apps to the users to increase the reputation level. This proposed research methodology provides a way for increasing the reputation level of the mobile owners by detecting and eliminating the fraudulent mobile apps.