Imperial College London
Portrait Picture

Professor Hamed Haddadi

Faculty of Engineering, Department of Computing

Professor of Human-Centred Systems
 
 
 
//

Contact

 

h.haddadi Website

 
 
//

Location

 

2Translation & Innovation Hub BuildingWhite City Campus

//

Summary

 

Publications

Citation

BibTex format

@article{Smith:2022:10.56553/popets-2022-0096,
author = {Smith, M and Snyder, P and Haller, M and Livshits, B and Stefan, D and Haddadi, H},
doi = {10.56553/popets-2022-0096},
journal = {Proceedings on Privacy Enhancing Technologies},
pages = {6--23},
title = {Blocked or broken? automatically detecting when privacy interventions break websites},
url = {http://dx.doi.org/10.56553/popets-2022-0096},
volume = {2022},
year = {2022}
}
Download

RIS format (EndNote, RefMan)

TY  - JOUR
AB  - A core problem in the development and maintenance of crowd-sourced filterlists is that their maintainers cannot confidently predict whether (and where)a new filter list rule will break websites. This is a result of enormity of theWeb, which prevents filter list authors from broadly understanding the impactof a new blocking rule before they ship it to millions of users. The inabilityof filter list authors to evaluate the Web compatibility impact of a new rulebefore shipping it severely reduces the benefits of filter-list-based contentblocking: filter lists are both overly-conservative (i.e. rules are tailorednarrowly to reduce the risk of breaking things) and error-prone (i.e. blockingtools still break large numbers of sites). To scale to the size and scope ofthe Web, filter list authors need an automated system to detect when a newfilter rule breaks websites, before that breakage has a chance to make it toend users.  In this work, we design and implement the first automated system forpredicting when a filter list rule breaks a website. We build a classifier,trained on a dataset generated by a combination of compatibility data from theEasyList project and novel browser instrumentation, and find it is accurate topractical levels (AUC 0.88). Our open source system requires no humaninteraction when assessing the compatibility risk of a proposed privacyintervention. We also present the 40 page behaviors that most predict breakagein observed websites.
AU  - Smith,M
AU  - Snyder,P
AU  - Haller,M
AU  - Livshits,B
AU  - Stefan,D
AU  - Haddadi,H
DO  - 10.56553/popets-2022-0096
EP  - 23
PY  - 2022///
SN  - 2299-0984
SP  - 6
TI  - Blocked or broken? automatically detecting when privacy interventions break websites
T2  - Proceedings on Privacy Enhancing Technologies
UR  - http://dx.doi.org/10.56553/popets-2022-0096
UR  - http://arxiv.org/abs/2203.03528v2
UR  - https://petsymposium.org/popets/2022/popets-2022-0096.php
UR  - http://hdl.handle.net/10044/1/105216
VL  - 2022
ER  -
Download