Fraud Application Detection Using Data Mining

 

Objective

The primary objective is to develop a system that finds ranking, rating and review behaviors for examining suggestions and then aggregation based on optimization to combine all the recommendations for detection of fraud.

Project Overview

Due to the fast growth of usage of mobile devices, mobile apps are essential in day-to-day activities of most of the people. Ranking and identifying the fraud is the critical challenge in front of the mobile App market because there is a large number of mobile Apps. App developers are using delicate means more and more frequently for increasing their Apps sales or posting fake App ratings. So it is necessary to prevent ranking fraud. This project introduces a system for mobile apps to rank fraud detection. The proposed method mines the leading sessions of mobile apps to precisely locate the ranking fraud. Furthermore, the system finds ranking, rating and review behaviors and investigation of three types of suggestion; they are ranking based suggestion, rating based suggestion and survey based suggestion is done. Then, an aggregation method based on optimization to combine all the suggestion for fraud detection is proposed. The system measure with App data collected from the App Store for an extended period.

Proposed System

Fraudulent Apps must be detected, as there is an increase in the number of mobile apps. This project aim is practical algorithm for identifying the leading sessions of each App based on its historical ranking of records. With the analysis of ranking behaviors of Apps, this system recognizes that the fraudulent Apps often has different ranking patterns in their every leading session compared with usual Apps. Some fraud suggestion identifies from Apps historical ranking records resulting in the development of three functions to detect likewise ranking based fraud suggestion. Moreover, two types of fraud suggestion based on Apps rating and review history are proposed.

This project represents the new novel approach for the development of a ranking fraud detection system for mobile apps. Initially, identification of rating based suggestion is done. Then identification of review based suggestion then by leading mining sessions ranking fraud suggestion is collected. And finally,the system performs the aggregation of all three suggestion to detect fraud apps. This method will offer considerable benefits and provides an opportunity to prevent fraudulent apps in the market.The important modules include,

• Rating Based Suggestions
• Review Based Suggestions
• Ranking Based Suggestions
• Aggregation of suggestion

Pre-processing of ratings: Ratings are between one to five, in this module, it will consider, the score which is less than or two are considered as worst, three as average and above three as best ratings. Pre processing reviews consists of tokenization, stop word removal and stemming. This new method called aggregation method combines all the three suggestions to detect the fraud. Rapid Miner is used here in this project to identify fraud app using data mining and sentiment analysis.

Software Requirements

• Windows OS
• Rapid Miner

Hardware Requirements

• Hard Disk – 1 TB or Above
• RAM required – 8 GB or Above
• Processor – Core i3 or Above

Technology Used

• Recommender System
• Data Mining
• Sentiment Analysis

Download Project

Download Abstract