AndroCT: Ten Years of App Call Traces in Android

A large-scale dataset on the dynamic profiles based on function calls of 35,974 benign and malicious Android apps from 10 historical years (2010 through 2019). Function calls are a commonly used means to model program behaviors, which may contribute to various code analysis approaches to assuring software correctness, reliability, and security. In particular, our dataset includes dynamic profiles of each app resulting from the same-length of time (10 mins) of being exercised by randomly generated inputs on both emulator and real device, enabling interesting and useful app analysis that reason about app behaviors in an evolutionary perspective while informing the differences of app behaviors on different run-time hardware platforms. Since we have 20 yearly datasets associated with 35,974 unique Android apps across the 10 years, profiling these apps took 12,000 hours. Considering the costs of filtering out apps that were originally sampled but that we were unable to profile (due to various reasons such as broken APKs, not being executable because of incompatibility issues, not instrumentable, etc.), we took over two years to produce all these traces. We hope to save future researchers' time in producing such a set of dynamic data to enable their empirical and technical work. 

==================

Thanks for your interest in our dataset. Collecting this dataset took tremendous computational and human effort. Thus, please observe the following restrictions in using our dataset: 

   -  Do not redistribute this dataset without our consent.
   -  Do not make commercial usage of this dataset.
   -  Get a faculty, or someone in a permanent position, to agree and commit to these conditions.
   -  When publishing your work that uses our dataset, please cite the following MSR 2021 data paper.

@inproceedings{AndroidCT,
  title = {AndroCT: Ten Years of App Call Traces in Android},
  author = {Wen Li, Xiaoqin Fu, and Haipeng Cai},
  booktitle = {The 18th International Conference on Mining Software Repositories (MSR 2021), Data Showcase Track},
  year = {2021},
}