Automated Detection of Doxing on Twitter

Abstract

Doxing refers to the practice of disclosing sensitive personal information about a person without their consent. This form of cyberbullying is an unpleasant and sometimes dangerous phenomenon for online social networks. Although prior work exists on automated identification of other types of cyberbullying, a need exists for methods capable of detecting doxing on Twitter specifically. We propose and evaluate a set of approaches for automatically detecting second-and third-party disclosures on Twitter of sensitive private information, a subset of which constitutes doxing. We summarize our findings of common intentions behind doxing episodes and compare nine different approaches for automated detection based on string-matching and one-hot encoded heuristics, as well as word and contextualized string embedding representations of tweets. We identify an approach providing 96.86% accuracy and 97.37% recall using contextualized string embeddings and conclude by discussing the practicality of our proposed methods.CCS Concepts: • Security and privacy → Privacy protections; Social aspects of security and privacy.

Introduction to the Minitrack on Value, Success, and Performance Measurements of Knowledge, Innovation and Entrepreneurial Systems

“…Publications in computer science have concentrated on the digital technologies that facilitate doxing. These papers have proposed designs to detect doxing on social media platforms (Karimi et al, 2022;Snyder et al, 2017) and analyzed internet technologies and software used for doxing (Khanna et al, 2016). In comparison, publications in social sciences and other disciplines have covered diverse topics.…”

“…As the effect on doxing victims might form part of a doxer's motivation, the characteristics of the doxer and doxing target have a joint effect in determining doxing behaviors. (Demydova, 2021) or as a negative byproduct of deliberative digital democracy (Buozis, 2019) Doxing of target groups: gendered online abuse that targets women, minorities (Eckert & Metzger-Riftkin, 2020), and white supremacists (Carriere et al, 2018) Doxing as a political tool to a candidate (Hansen & Lim, 2019) Law and regulation Critical assessments of current regulation: analyzing the limitations of current statutes (Amiruddin et al, 2021;Lindvall, 2019;MacAllister, 2016) Application of current legislation: appeal to regulate doxing from the victim's rights to privacy (Bei Li, 2018;Calabro, 2018;Corbridge, 2018) or the "right to be forgotten" (Pittman, 2018;Yudiana et al, 2022), as a form of cyber-harassment (Mery, 2020), and the intentional infliction of emotional distress (McIntyre, 2016) Develop novel forms of regulation: proposing to allocate regulatory responsibilities to organizations for institutional doxing (Styple, 2021) and to expend additional social resources for regulation (Crompton, 2018) Technology Proposing doxing detection methods based on string-matching and encoded heuristics (Karimi et al, 2022;Snyder et al, 2017) Analyzing digital software (Maltego) that gathers information from open sources and thus facilitates doxing (Khanna et al, 2016)…”

“…Although IT applications have facilitated doxing and had a negative influence, they also have the potential to constrain doxing. Karimi et al (2022) and Snyder et al (2017) proposed doxing detection methods on Twitter, 4chan.org, and pastebin.com that can identify doxing information from available online information. However, further studies are needed to test the practicality of these detection methods.…”

Addressing Resource and Privacy Constraints in Semantic Parsing Through Data Augmentation

“…Finally, the detection of PII in user data is an applied topic of interest (Pilán et al, 2022), such as for safely summarizing call transcripts (Transcribe) or the automatic detection of doxing (Karimi et al, 2022). In our work we implement a solution meant as a proof of concept for our exploration, based on detecting and replacing named entities.…”