GenAI4SG @ CIKM 2026

About The Workshop

Building useful GenAI where the stakes are real.

Generative AI has become a general-purpose technology for producing text, images, code, and explanations across knowledge-intensive tasks. This creates new opportunities for social good in domains where expertise, data, and resources are scarce, including rare diseases, low-resource education, accessibility, crisis response, sustainability, public-service access, and information integrity.

These are also the settings where failures can be most harmful. GenAI systems may hallucinate, reproduce linguistic or cultural biases, privilege high-resource languages, leak sensitive information, amplify inequalities, or impose computational costs that make deployment impractical for low-resource organizations. GenAI4SG brings together researchers working on retrieval-augmented generation, agentic systems, multimodal models, and knowledge-grounded AI to study how these technologies can be responsibly designed, evaluated, and deployed in line with the UN Sustainable Development Goals.

The workshop is aligned with the CIKM community, drawing on information retrieval, knowledge management, databases, data mining, machine learning, NLP, and recommender systems to retrieve evidence, build and maintain knowledge bases, generate grounded explanations, evaluate information quality, and deploy systems over heterogeneous data.

Held with the 35th ACM International Conference on Information and Knowledge Management, November 7-11, 2026.

September 13, 2026 Paper submission deadline

September 23, 2026 Recommended acceptance notification

TBD Camera-ready papers, archival track

November 8, 2026 Workshop day at CIKM

Call For Contributions

Topics of interest

We welcome technical, empirical, resource, demo, and position papers on methods, systems, datasets, benchmarks, and applications.

Retrieval-augmented generation and evidence-grounded GenAI for high-impact domains.
Knowledge-graph integration, semantic grounding, fact-checking, and source verification.
LLM agents and agentic workflows for social-good applications.
Multimodal GenAI for health, education, accessibility, crisis response, and sustainability.
Synthetic data generation, data-centric GenAI, and low-resource adaptation.
Benchmarks, datasets, and evaluation protocols for societal benefit and deployment risk.
Fairness, robustness, privacy, explainability, and human-centered assessment.
Efficient, sustainable, and practical deployment for public-interest organizations.
Applications supporting vulnerable and underserved populations.
Information integrity, misinformation detection, and coordinated manipulation analysis.

Submission

Two submission categories

Authors decide whether their paper is archival or non-archival upon acceptance, when uploading the camera-ready version. Both categories are reviewed by the same program committee, under the same criteria, and both are presented at the workshop if accepted.

Archival

Published in the proceedings

For authors who want a citable publication of the work as presented.

Included in the workshop proceedings, which shall be submitted to CEUR-WS.org for online publication.
Open access under the Creative Commons Attribution 4.0 International license, with authors keeping copyright.
Camera-ready papers use the CEURART template and include the required Declaration on Generative AI.
The contact author returns a hand-signed author agreement, since scanned pen-on-paper signatures are the only form accepted.

Non-archival

Presented, not published

For authors who intend to submit an extended version elsewhere, or who are bound by another venue's dual-submission rules.

Reviewed and presented in the same sessions as archival papers.
Not included in the proceedings, so the work remains free for submission to a journal or a major conference.
No template or licensing requirements beyond the review format.
Listed in the program as a presentation without an accompanying published paper.

The category is selected at camera-ready time, after acceptance, and cannot be changed afterwards. Reviewers are not told which category a paper belongs to, and the preface of the proceedings will report the number of submissions and acceptances in each category.

Guidelines

Submissions are handled electronically through EasyChair and are reviewed in the ACM two-column format. Accepted archival papers are then reformatted into the single-column CEURART template for the camera-ready version.

Full papers are 8 to 12 pages and present mature technical or empirical contributions. Short papers and position papers are 4 to 6 pages and present preliminary results or research agendas. Dataset, benchmark, and demo papers presenting reusable resources or systems may use either length. References do not count towards the page limits.

Each submission will receive at least two reviews. Evaluation criteria include relevance to the workshop, technical quality, novelty, clarity, methodological soundness, potential societal benefit, and awareness of risks for vulnerable or underserved populations.

Participation is open to all registered CIKM attendees. At least one author of each accepted paper is expected to register and present the work on-site in Rome.

Submit via EasyChair

Important Notes

Workshop papers are not included in the ACM CIKM proceedings. Proceedings shall be submitted to CEUR-WS.org for online publication, and acceptance of the volume is at the discretion of the CEUR-WS team.

Papers already published elsewhere cannot be included in the volume. Authors of archival papers keep the right to post the published version on their homepage or institutional repository.

All deadlines are 11:59 PM Anywhere on Earth unless stated otherwise.

ACM Template, For Review CEURART, For Camera-Ready

Planned Special Issue

Generative AI for Vulnerable and Underserved Populations

Information Processing & Management, Elsevier

A special issue on this theme is in preparation with Information Processing & Management, covering grounding, evaluation, and deployment for populations that have no easy way to verify the information a generative system gives them. Topics include retrieval-augmented generation over thin or contested evidence bases, knowledge base construction and maintenance, verification and information integrity, evaluation protocols that measure deployment risk rather than benchmark performance, and efficient deployment under real resource constraints.

The call will be an open one, not a proceedings volume, and the majority of submissions are expected to come from outside the workshop. Authors of accepted workshop papers, in either category, will be invited to submit substantially extended versions. All submissions go through the journal's standard peer review, and an invitation carries no guarantee of acceptance.

The full call for papers, including scope, deadlines, and guest editors, will be published here once the special issue is formally established with the journal.

Program

Tentative half-day schedule

Final times and accepted papers will be announced after the submission and review process. The published program will indicate which presentations correspond to papers included in the proceedings and which do not.

5 min

Opening

Workshop motivation and overview.

45 min

Invited Talk

Given by Donato Pasqualicchio (Microsoft).

40 min

Paper Session I

Methods and systems, including RAG, knowledge grounding, multimodal models, agents, synthetic data, and efficient deployment, followed by joint panel Q&A.

30 min

Coffee Break and Interaction

Informal discussion, posters, and demo interaction.

40 min

Paper Session II

Applications, evaluation, and deployment in health, education, accessibility, crisis response, sustainability, and information integrity, followed by joint panel Q&A.

20 min

Open Discussion and Closing

Open challenges for GenAI systems serving vulnerable and underserved populations.

Invited Speaker

Donato Pasqualicchio, Microsoft

The talk title and abstract will be updated here as soon as they are confirmed.

Donato Pasqualicchio

Azure Solution Specialist

Talk title

To be announced. The title of the talk will be published here once it is finalized.

Abstract

To be announced. A short abstract will be added ahead of the workshop.

Biography

Donato Pasqualicchio has spent the past 25 years making technology useful and, occasionally, stopping it from making life more complicated. An Azure Solution Specialist at Microsoft and former Senior Cloud Solution Architect, he helps organisations turn ambitious cloud and AI ideas into solutions that work in the real world.

He has been working with AI since the days when most people looked at it like cows watching a passing train: puzzled, slightly suspicious, and unsure why it mattered. His experience spans machine learning, NLP, computer vision, conversational systems, and generative AI.

Donato was among the contributors who shaped Microsoft’s Responsible AI strategy and has spoken at numerous Microsoft events about AI, cloud transformation, governance, and responsible innovation.

He believes in “productive laziness”: automate repetitive work, reuse what works, and save human creativity for problems that genuinely deserve it.

The invited talk is scheduled for 45 minutes, including Q&A.

Organizing Committee

Organizers

Martino Ciaperoni

Postdoctoral Research Scholar Scuola Normale Superiore

His work spans data mining, data management, and trustworthy AI, currently focusing on biomedical applications and explainability of transformer-based language models.

Marco Minici

Research Scientist ICAR-CNR, National Research Council of Italy

His work focuses on online social safety, social computing, algorithmic bias, and ML methods for detecting coordinated information operations.

Vincenzo Moscato

Full Professor University of Naples Federico II

Full Professor of Information Systems, coordinator of the MSc in Data Science, and Director of the CINI ITEM national laboratory.

Marco Postiglione

Postdoctoral Research Scholar Northwestern University

His work bridges theoretical advances and deployed, trustworthy AI for social good that serves real users, from fact-checkers and journalists to clinicians and public-safety organizations.

Current Program Committee · Still Growing

Reviewers

Valerio La Gatta, Northwestern University
Eliana Pastor, Politecnico di Torino
Mariano Barone, University of Naples Federico II
Giuseppe Riccio, University of Naples Federico II
Aakanksha Joshi, IBM

Generative AI for Social Good and Vulnerable Populations