Alireza Mohammadinodooshan

arXiv2025

Public versus Less-Public News Engagement on Facebook: Patterns Across Bias and Reliability

The rapid growth of social media as a news platform has raised significant concerns about the influence and societal impact of biased and unreliable news on these platforms. While much research has explored user engagement with news on platforms like Facebook, most studies have focused on publicly shared posts. This focus leaves an important question unanswered: how representative is the public sphere of Facebook's entire ecosystem? Specifically, how much of the interactions occur in less-public spaces, and do public engagement patterns for different news classes (e.g., reliable vs. unreliable) generalize to the broader Facebook ecosystem? This paper presents the first comprehensive comparison of interaction patterns between Facebook's more public sphere (referred to as public in paper) and the less public sphere (referred to as private). For the analysis, we first collect two complementary datasets: (1) aggregated interaction data for all Facebook posts (public + private) for 19,050 manually labeled news articles (225.3M user interactions), and (2) a subset containing only interactions with public posts (70.4M interactions). Then, through discussions and iterative feedback from the CrowdTangle team, we develop a robust method for fair comparison between these datasets. Our analysis reveals that only 31% of news interactions occur in the public sphere, with significant variations across news classes. Engagement patterns in less-public spaces often differ, with users, for example, engaging more deeply in private contexts. These findings highlight the need to examine both public and less-public engagement to fully understand news dissemination on Facebook. The observed differences hold important implications on content moderation, platform governance, and policymaking, contributing to healthier online discourse.

cs.SI↗

arXiv2020

Comment on "AndrODet: An adaptive Android obfuscation detector"

We have identified a methodological problem in the empirical evaluation of the string encryption detection capabilities of the AndrODet system described by Mirzaei et al. in the recent paper "AndrODet: An adaptive Android obfuscation detector". The accuracy of string encryption detection is evaluated using samples from the AMD and PraGuard malware datasets. However, the authors failed to account for the fact that many of the AMD samples are highly similar due to the fact that they come from the same malware family. This introduces a risk that a machine learning system trained on these samples could fail to learn a generalizable model for string encryption detection, and might instead learn to classify samples based on characteristics of each malware family. Our own evaluation strongly indicates that the reported high accuracy of AndrODet's string encryption detection is indeed due to this phenomenon. When we evaluated AndrODet, we found that when we ensured that samples from the same family never appeared in both training and testing data, the accuracy dropped to around 50%. Moreover, the PraGuard dataset is not suitable for evaluating a static string encryption detector such as AndrODet, since the particular obfuscation tool used to produce the dataset effectively makes it impossible to extract meaningful features of static strings in Android apps.

cs.CR↗