Malware Detection and Classification Using Latent Semantic Indexing

  • Srijana Parajuli Department of Electronics and Computer Engineering, Pulchowk Campus, T.U.
  • Subarna Shakya Department of Electronics and Computer Engineering, Pulchowk Campus, T.U.
Keywords: android, intent, permission, weight score, malware detection, latent semantic indexing

Abstract

The increasing popularity of smart phones has led to the dramatic growth in mobile malware especially in Android platform.  Many aspects of android permission has been studied for malware detection but sufficient attention has not been given to intent.  This  research  work  proposes  using  Latent  Semantic  Indexing  for  malware  detection  and  classification  with  permissions and intents based features. This method analyses the Manifest file of an android application by understanding the risk level of permission and intents and assigning weight score based on their sensitivity. In an experiment conducted using a dataset containing 400 malware samples and 400 normal/benign samples, the results show accuracy of 83.5% using Android Intent against 79.1 % using Android permission. Additionally, experiment on combination of both features results in accuracy of 89.7%. It can be concluded from this research work that dataset with intent based features is able to detect malwares more when compared to permissions based features.

Downloads

Download data is not yet available.
Abstract
1931
pdf
260
Published
2018-12-31
How to Cite
Parajuli, S., & Shakya, S. (2018). Malware Detection and Classification Using Latent Semantic Indexing. Journal of Advanced College of Engineering and Management, 4, 153-161. https://doi.org/10.3126/jacem.v4i0.23205
Section
Articles