<P>Android application (app) stores contain a<I> huge</I> number of apps, which are<I> manually</I> classified based on the apps’ descriptions into various categories. However, the predefined categories or apps desc...
http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
https://www.riss.kr/link?id=A107697969
2018
-
SCOPUS
학술저널
1-21(21쪽)
0
상세조회0
다운로드다국어 초록 (Multilingual Abstract)
<P>Android application (app) stores contain a<I> huge</I> number of apps, which are<I> manually</I> classified based on the apps’ descriptions into various categories. However, the predefined categories or apps desc...
<P>Android application (app) stores contain a<I> huge</I> number of apps, which are<I> manually</I> classified based on the apps’ descriptions into various categories. However, the predefined categories or apps descriptions are usually<I> not</I> very accurate to reflect the real functionalities of apps, thereby leading to<I> misclassify</I> the apps, which may cause serious<I> security issues</I> and<I> unreliability</I> problem in the app store. Therefore, the automatic app classification is an<I> important</I> demand to construct a<I> secure</I>,<I> reliable</I>,<I> integrated</I>, and<I> easy to navigate</I> app store. In this paper, we propose an effective method called<I> AndroClass</I> to<I> automatically</I> classify apps based on their<I> real</I> functionalities by using<I> rich</I> and<I> comprehensive</I> features representing the<I> actual</I> functionalities of the apps. AndroClass performs<I> three</I> steps of<I> feature extraction</I>,<I> feature refinement</I>, and<I> classification</I>. In the feature extraction step, we extract 14 various features for each app by utilizing a<I> unified tool suite</I>. In the feature refinement step, we apply<I> Random Forest</I> algorithm to refine the features. In the classification step, we combine refined features into a<I> single</I> one and AndroClass is equipped with K-Nearest Neighbor, Naive Bayes, Support Vector Machine, and Deep Neural Network to classify apps. On the contrary to the existing methods, all the utilized features in AndroClass are<I> stable</I> and<I> clearly</I> represent the actual functionalities of the app, AndroClass does<I> not</I> pose any issues to the<I> user privacy</I>, and our method can be applied to classify<I> unreleased</I> or<I> newly released</I> apps. The results of<I> extensive</I> experiments with two<I> real-world</I> datasets and a dataset constructed by<I> human experts</I> demonstrate the effectiveness of AndroClass where the classification accuracy of AndroClass with the latter dataset is 83.5%.</P>
MC-GiV2V: Multichannel Allocation in mmWave-Based Vehicular Ad Hoc Networks
Modulation Set Optimization for the Improved Complex Quadrature SM