Protecting Your Children from Inappropriate Content in Mobile Apps: An Automatic Maturity Rating Framework

摘要: Mobile applications (Apps) could expose children or adolescents to mature themes such as sexual content, violence and drug use, which results in an inappropriate security privacy risk for them. Therefore, mobile platforms provide rating policies label the maturity levels of Apps reasons why App has a given level, enables parents select maturity-appropriate their children. However, existing approaches implement these are either costly (because expensive manually labeling) inaccurate no centralized controls). In this work, we aim design build machine learning framework automatically predict associated with high accuracy low cost. To end, take multi-label classification approach contents then level according policy. Specifically, extract novel features from descriptions by leveraging deep technique capture semantic similarity pairwise words adapt Support Vector Machine correlations pearson correlation setting. Moreover, evaluate our various baseline methods using datasets that collected both Store Google Play. We demonstrate that, only descriptions, already achieves 85% Precision predicting 79% levels, substantially outperforms methods.