We address the problem of identifying in-app user actions from Web access logs when the content of those logs is both encrypted (through HTTPS) and also contains automated Web accesses. We nd that the distribution of time gaps between HTTPS accesses can dis- tinguish user actions from automated Web accesses generated by the apps, and we determine that it is reasonable to identify meaningful user actions within mobile Web logs by modelling this temporal feature. A real-world experiment is conducted with multiple mobile devices running some popular apps, and the results show that the proposed clustering- based method achieves good accuracy in identifying user actions, and outperforms the state-of-the-art baseline by 17:84%.
@inproceedings{DBLP:conf/pakdd/PriyogiSSCTR18,
author = {Bilih Priyogi and
Mark Sanderson and
Flora D. Salim and
Jeffrey Chan and
Martin Tomko and
Yongli Ren},
bibsource = {dblp computer science bibliography, https://dblp.org},
biburl = {https://dblp.org/rec/conf/pakdd/PriyogiSSCTR18.bib},
booktitle = {Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia
Conference, PAKDD 2018, Melbourne, VIC, Australia, June 3-6, 2018,
Proceedings, Part II},
doi = {10.1007/978-3-319-93037-4_24},
editor = {Dinh Q. Phung and
Vincent S. Tseng and
Geoffrey I. Webb and
Bao Ho and
Mohadeseh Ganji and
Lida Rashidi},
pages = {300--311},
publisher = {Springer},
series = {Lecture Notes in Computer Science},
timestamp = {Sat, 19 Oct 2019 01:00:00 +0200},
title = {Identifying In-App User Actions from Mobile Web Logs},
url = {https://doi.org/10.1007/978-3-319-93037-4_24},
volume = {10938},
year = {2018}
}
© 2021 Flora Salim - CRUISE Research Group.