Social & Information Networks

Daniel Miehling, Sandra Kuebler

This work addresses the limited research on short-form video news for humanities scholars, though it is incremental as it applies existing methods to a new data format.

Jonathan Stray, Ian Baker, George Beknazar-Yuzbashev et al. · uw

This addresses the societal issue of polarization on social media for users and platforms, but it is incremental as it builds on existing algorithmic interventions.

Eleanor A. Power, Monique Borgerhoff Mulder, Samuel Bowles et al.

This research provides empirical evidence for the link between social network structure and wealth inequality across a diverse range of human societies, addressing a gap in previous studies that relied primarily on online social media data.

Sahajpreet Singh, Insyirah Mujtahid, Min-Yen Kan et al.

For researchers and practitioners of misinformation detection, this benchmark provides a more realistic evaluation of LLMs in public, multilingual settings, highlighting critical gaps in source selection that need addressing.

Sebastian Pohl, Harsh Mehta, Pranav Mambayil et al.

This paper provides a rigorous evaluation of LLMs as proxies for human participants in social science experiments, revealing critical limitations that affect researchers relying on such simulations.

Roben Delos Reyes, Timothy Douglas, Asanobu Kitamoto

This provides a scalable alternative to real-world data curation for crisis informatics researchers, though it is incremental as it adapts existing synthetic data methods to a specific domain.

Xinnong Zhang, Wanting Shan, Hanjia Lyu et al.

For researchers and practitioners using LLMs to simulate online opinion dynamics, this work provides an evaluation framework to assess context sensitivity, highlighting both promise and risk.

Renhong Huang, Ning Tang, Jiarong Xu et al.

This addresses the societal risk of delayed and costly policy evaluation for social platforms, offering a pre-deployment alternative to reactive A/B testing.

Siqi Miao, Ziyang Chen, Yuhong Luo et al.

For researchers studying network dynamics and cybersecurity, this work addresses the gap in LLM-based simulations for domains requiring realistic network topologies.

Soorya Ram Shimgekar, Vipin Gunda, Jiwon Kim et al.

This addresses a safety concern for vulnerable users interacting with AI, providing empirical evidence on a previously anecdotal phenomenon, though it is incremental in quantifying and mitigating the effect.

Prince Zizhuang Wang, Shuli Jiang

This addresses privacy vulnerabilities in emerging AI agent networks for users, though it is incremental as it builds on prior multi-agent and privacy work.

Ali Al-Lawati, Nafis Tripto, Abolfazl Ansari et al.

For developers and moderators of multi-agent systems, this addresses the challenge of detecting malicious agents that evade content-based filters by focusing on intent rather than surface-level content.

Zhi Zeng, Yifei Yang, Jiaying Wu et al.

This addresses the need for robust and explainable detection of micro-video misinformation, which impacts public trust, though it is incremental in improving benchmarks and methods.

Anjun Hu, Hanting Xie, Saranya Govindan et al.

For researchers and practitioners building LLM-powered multi-agent recommendation systems, this work provides initial insights into how connectivity affects vulnerability, but it is an incremental adaptation with limited practical guidance.

Chenxu Zhu, Hantao Yao, Wu Liu et al.

For researchers studying geopolitical opinion dynamics, this provides a more realistic simulation framework that captures event-driven shifts, though it is domain-specific.

Yechao Zhang, Shiqian Zhao, Jie Zhang et al.

This addresses a critical security problem for users of personal AI agents, revealing an inherent architectural flaw that enables silent memory pollution without requiring prompt injection.

Vageesh Kumar Saxena

For law enforcement, this work offers a data-driven method to connect anonymous online profiles involved in human trafficking and illicit trade, though the approach is incremental.

Zihang Fu, Fanxiao Li, Jianyang Gu et al.

For health misinformation governance on social platforms, EvoNote provides a scalable, evidence-grounded correction method that outperforms human-written notes in quality and speed.

Xinze Li, Nanyun Peng, Simone Severini et al.

For developers of formal mathematics libraries, this work quantifies structural inefficiencies and mismatches between human-designed taxonomies and logical dependencies.

Junsuk Rho, Jinn-Kong Sheu, Andrew Forbes et al.

For the research community, this perspective provides a framework and actionable recommendations to reform evaluation practices, but it is a conceptual contribution without empirical validation.