A Machine Learning Based Approach To App Rating Manipulation Detection

Open Access
Author:
Song, Yang
Graduate Program:
Computer Science and Engineering
Degree:
Master of Science
Document Type:
Master Thesis
Date of Defense:
April 02, 2014
Committee Members:
  • Sencun Zhu, Thesis Advisor
Keywords:
  • machine learning
  • app rating manipulation
  • app store
Abstract:
In order to promote apps in mobile app stores, for malicious developers and users, manipulating average rating is a popular and feasible way. In this thesis, we propose a two-phase machine learning approach to detect the app rating manipulation. We give the definitions of abused app, malicious user and collusion group, and characteristics. In the first learning phase, we generate feature ranks for different app stores and find that top features match the characteristics of abused apps and malicious users. In the second learning phase, we choose top N features and train our models for each app store. With cross-validation, our training models achieve 85% f-score. We also use our training models to discover new suspicious apps from our data set and evaluate them with two criteria. Finally, we conduct some analysis based on the suspicious apps classified by our training models and some interesting results are discovered.