Description
The rapid proliferation of Large Language Models, autonomous web agents, and AI-driven data pipelines has fundamentally transformed how information is created, retrieved, and consumed on the Web. While these advances unlock unprecedented capabilities, from intelligent knowledge extraction to conversational interfaces for complex databases, they also introduce serious challenges around transparency, accountability, fairness, and societal impact. As the Web continues to evolve into the primary substrate connecting people, institutions, and intelligent machines, ensuring that AI systems operating at web scale are both trustworthy and genuinely human-centered has become one of the most pressing research imperatives of our time.
This special track invites original, high-quality contributions that sit at the intersection of Web information systems, responsible AI, and human factors. We are particularly interested in work that goes beyond raw performance benchmarks to critically examine how AI-powered web systems behave under adversarial conditions, how they can be made interpretable and auditable by non-expert users, and how they affect diverse communities across different cultural and socioeconomic contexts. Submissions may address foundational research questions — such as the theoretical underpinnings of trustworthy LLM-based web retrieval or the formal verification of web agents, as well as applied and experimental work targeting real-world platforms in e-commerce, e-government, e-learning, social networks, and beyond.
The track takes an explicitly interdisciplinary stance, welcoming contributions from researchers working in databases, information retrieval, machine learning, human-computer interaction, security, and the social sciences. This breadth reflects a core conviction: no single discipline can adequately address the full complexity of responsible AI deployment on the Web. We particularly encourage submissions from Asia Pacific regions and other underrepresented communities, as diverse perspectives are essential for building web systems that serve humanity equitably.
Topics of Interest
The track welcomes contributions on the following topics, though the list is not exhaustive:
- Trustworthy and interpretable AI for web-scale data management, including explainability methods for LLM-based retrieval and recommendation systems, auditing frameworks for automated web pipelines, and formal approaches to AI reliability on the Web.
- LLM-augmented web information systems, covering retrieval-augmented generation (RAG) architectures, LLM-based query processing and optimization over web-scale knowledge bases, hallucination detection and mitigation in web-facing AI applications, and evaluation methodologies for LLM-driven web agents.
- Human factors, fairness, and social impact on the Web, including algorithmic fairness in web search and advertising, bias detection in web-mined datasets, accessibility of AI-driven web interfaces, and the societal consequences of automated content generation and curation.
- Web security, privacy, and adversarial robustness, encompassing adversarial attacks on web-based AI models, data poisoning in web crawling and training pipelines, privacy-preserving web analytics and federated learning for web data, and trust management in distributed web service architectures.
- Responsible knowledge extraction and management, addressing ethically-aware information extraction from web corpora, provenance tracking in linked open data and knowledge graphs, transparent semantic web and ontology construction, and the governance of web-scale knowledge systems.
- Human-centered rich interfaces and web intelligence, including HCI design principles for AI-assisted web applications, user modeling under uncertainty, explainable web recommendations, and participatory design methodologies for web information systems.
- Crowdsourcing, social web, and collective intelligence under scrutiny, covering quality assurance in crowdsourced web data, misinformation detection and provenance in social networks, responsible web community analysis, and transparent web-based collaborative systems.
- Evaluation, benchmarking, and reproducibility, specifically work that proposes new datasets, metrics, or experimental protocols designed to assess the trustworthiness, robustness, and human-centeredness of AI systems operating on web-scale information.
Submission Guidelines
All submissions must be original and must not be under review elsewhere at the time of submission. Papers should be formatted according to the main conference guidelines and submitted through the conference management system via the dedicated special track submission page. Both full research papers (up to 15 pages) and short papers presenting early-stage or position work (up to 8 pages) are welcome.
