Filter List Generation for Underserved Regions

Abstract

Filter lists play a large and growing role in protecting and assisting web users. The vast majority of popular filter lists are crowd-sourced, where a large number of people manually label resources related to undesirable web resources (e.g. ads, trackers, paywall libraries), so that they can be blocked by browsers and extensions.Because only a small percentage of web users participate in the generation of filter lists, a crowd-sourcing strategy works well for blocking either uncommon resources that appear on "popular" websites, or resources that appear on a large number of "unpopular" websites. A crowd-sourcing strategy will perform poorly for parts of the web with small "crowds", such as regions of the web serving languages with (relatively) few speakers.This work addresses this problem through the combination of two novel techniques: (i) deep browser instrumentation that allows for the accurate generation of request chains, in a way that is robust in situations that confuse existing measurement techniques, and (ii) an ad classifier that uniquely combines perceptual and page-context features to remain accurate across multiple languages.We apply our unique two-step filter list generation pipeline to three regions of the web that currently have poorly maintained filter lists: Sri Lanka, Hungary, and Albania. We generate new filter lists that complement existing filter lists. Our complementary lists block an additional 3,349 of ad and ad-related resources (1,771 unique) when applied to 6,475 pages targeting these three regions.We hope that this work can be part of an increased effort at ensuring that the security, privacy, and performance benefits of web resource blocking can be shared with all users, and not only those in dominant linguistic or economic regions. CCS CONCEPTS• Security and privacy → Human and societal aspects of security and privacy; • Information systems → Web applications.

AutoFR: Automated Filter Rule Generation for Adblocking

“…One line of prior work aims to develop ML models to automatically generate filter rules for blocking ads [11,36,74]. Bhagavatula et al [11] (BD+) trained supervised ML classifiers to detect advertising URLs.…”

“…The research community is actively working on machine learning (ML) approaches to assist with filter rule generation [11,36,74] or to build models to replace filter lists altogether [1,41,73,89]. There are two key limitations of prior ML-based approaches.…”

WebGraph: Capturing Advertising and Tracking Information Flows for Robust Blocking

“…Graph-based approaches extract features from the crosslayer graph representation to train ML models to detect ads and trackers [18,19]. These approaches leverage rich crosslayer context and thus claim to be robust to evasion attempts.…”

“…It extracts structural features from the graph such as node connectivity and ancestry information as well as content features such as URL length and presence/absence of certain keywords. Sjösten et al [19] introduced PageGraph, which extends ADGRAPH's graph representation by improving event attribution and capturing more behaviors. In addition to content and structural features, they also added perceptual features to train the classifier.…”

“…However, these ML-based blocking approaches are highly susceptible to adversarial evasion techniques that are already found in the wild, including URL obfuscation [16] and code obfuscation [17]. To address this limitation, the next generation of ML-based blocking approaches leverage cross-layer graph information from multiple layers of the web stack [18,19]. These approaches claim better robustness to evasion, as compared to single-layer approaches, due to their use of structural features of the graph (i.e., the hierarchy of resource inclusions) in addition to traditional content features (i.e., the resource's network location or response content).…”