GENNEXT-SIGIR-25

Home

Large Language Models (LLMs) and other generative architectures are rapidly reshaping the fields of Information Retrieval (IR) and Recommender Systems (RS). Advanced language agents—which combine LLMs with specialized tool usage, multi-turn dialogue, and domain knowledge—unlock new capabilities such as context-sensitive retrieval, personalized generation, and adaptive conversation flows.

Despite the potential benefits, these generative models introduce new challenges in terms of hallucination, bias/fairness, data privacy, security, and evaluation methodologies. GENNEXT aims to explore the intersection of LLM-based language agents, generative content creation, and conversational AI in IR and RS—focusing on risks, opportunities, and novel forms of evaluation and user interaction.

Our workshop builds upon the success of the ROEGEN@RecSys'24 event but broadens the scope to include more general information retrieval, next-generation recommendation, and tool-augmented LLMs.


Call for Papers

We invite researchers and practitioners to submit work related (but not limited) to:

  • LLM-driven IR and Recommender Systems
    Prompting, in-context learning, foundation models, or domain-adaptive fine-tuning for IR/RS.
  • Agentic Tool Usage
    Techniques enabling LLMs to call external APIs (e.g., knowledge bases, retrieval models, recommendation engines) to fulfill user queries.
  • Conversational and Dialogue Systems
    Multi-turn interactions, user modeling, dynamic preference elicitation, explanation, or negotiation with generative AI.
  • Generative Content Creation
    Novel item generation (text, images, music), creative item repurposing, or integrated retrieval-plus-generation frameworks.
  • Bias and Fairness, Privacy, and Ethics
    Identifying and mitigating biases in generative models, ensuring privacy, building trust, and tackling hallucination risks.
  • Evaluation Metrics and Benchmarks
    Designing new metrics or user study protocols that capture the interplay of generative quality, recommendation relevance, and ethical concerns.

Submission Guidelines

  • Full Papers: Up to 9 pages (including figures, tables, proofs, appendixes, acknowledgments, and any content except references) but excluding references. Present substantial research, theoretical analyses, or comprehensive surveys.
  • Short Papers: Up to 4 pages (including figures, tables, proofs, appendixes, acknowledgments, and any content except references) but excluding references. Suitable for work-in-progress or preliminary findings.
  • Extended Abstracts: 2–3 pages (excluding references). For late-breaking results, vision papers, or discussion proposals.

Submissions must be anonymized and follow the official ACM two-column SIGCONF template.
All submissions will undergo double-blind peer review.

Accepted papers will be published in CEUR-WS or a similar open-access venue. Selected high-quality submissions may be invited for extension in a journal special issue.

Submission Link: EasyChair for GENNEXT@SIGIR'25


Important Dates

  • Submission Deadline: April 23, 2025 (AOE) New: May 4, 2025 (AOE)
  • Notification of Acceptance: May 21, 2025
  • Workshop Date: July 17, 2025 (During SIGIR 2025)
    Location: Mantegna 1, Floor 3

Exact deadlines may be adjusted to align with SIGIR final scheduling.


Program

Time Event
09:00-09:45 Invited Talk: ChengXiang Zhai (UIUC)
"Towards A Unified Agentic Framework for Conversational Information Retrieval and Recommendation: Models, Algorithms, and Evaluation"
09:45-10:30 Contributed Talks: "Resources, Evaluation, RAG, and Information Retrieval"
10:30-11:00 Coffee Break
11:00-11:45 Invited Talk: Julian McAuley (UCSD)
"What's still hard about conversational recommendation?"
11:45-12:33 Contributed Talks: "Generative Models for Conversation and Recommendation"
12:33-12:40 Wrap-up and Future Directions
13:00-14:30 Joint SIGIR Workshops Poster Session (poster instructions)

  • Keynote 1: ChengXiang Zhai (UIUC)
    Title: "Towards A Unified Agentic Framework for Conversational Information Retrieval and Recommendation: Models, Algorithms, and Evaluation"
    Abstract: Conversational Information Retrieval (IR) and Recommendation Systems have traditionally been treated as distinct yet closely related tasks. However, recent advancements in large language models (LLMs), interactive user modeling, and reinforcement learning suggest the possibility of viewing both IR and recommender systems as special cases within a broader agentic interaction framework. In this talk, I propose a unified perspective where conversational IR and recommendation tasks are modeled as iterative, cooperative interactions between intelligent agents and users. Key technical challenges and ideas will be discussed, including leveraging LLMs for building general-purpose conversational agents, designing effective memory mechanisms for context retention, modeling diverse user tasks and behaviors, and applying reinforcement learning for adaptive interaction management. Furthermore, I will explore evaluation methodologies, particularly emphasizing the use of simulated user agents to assess the effectiveness and adaptability of these unified conversational agents.
  • Keynote 2: Julian McAuley (UCSD)
    Title: "What's still hard about conversational recommendation?"
    Abstract: In this talk we'll explore the current landscape of conversational recommendation in light of new developments on Large Language Models. While LLMs offer a surprisingly effective "off-the-shelf" solution to conversational recommendation, they also have various limitations. We'll look at ways that current models can potentially be improved by exploring new datasets, methods, and evaluation protocols for conversational recommendation.

Workshop Organizers

  • Yashar Deldjoo, Tenure-Track Assistant Professor, Polytechnic University of Bari, Italy
  • Julian McAuley, Professor, UC San Diego, USA
  • Scott Sanner, Professor, University of Toronto, Canada
  • Pablo Castells, Professor, Autonomous University of Madrid, Spain
  • Shuai Zhang, Applied Scientist, Amazon Web Services AI, USA
  • Enrico Palumbo, Senior Research Scientist, Spotify
  • Hugues Bouchard, Senior Research Manager, Spotify

Program Committee

  • Yupeng Hou (UCSD)
  • Rishabh Mehrotra (Sharechat)
  • Chengkai Huang (University of New South Wales)
  • Mohamed Reda Bouadjenek (Deakin University)
  • José Luis Redondo García (Spotify)
  • Ali Vardasbi (University of Amsterdam)
  • Martin Mladenov (Google Research)
  • Branislav Kveton (Adobe Research)
  • Qidong Liu (City University of Hong Kong)
  • Azin Ghazimatin (Max Planck Institute)
  • Gustavo Penha (Spotify)

Contact and Further Information

For any inquiries, please email: gennext_at_sigir2025@googlegroups.com and deldjooy@acm.org

© 2025 GENNEXT@SIGIR. All rights reserved.