Modern Recommender Systems Leveraging Generative AI: Fundamentals, Challenges and Opportunities

Tutorial at 2024 ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Barcelona, Spain, 2024
Schedule: August 25, 2024

Abstract

Traditional recommender systems (RS) have used user-item rating histories as their primary data source, with collaborative filtering being one of the principal methods. However, generative models have recently developed abilities to model and sample from complex data distributions, including not only user-item interaction histories but also text, images, and videos - unlocking this rich data for novel recommendation tasks. Through this comprehensive and multi-disciplinary survey, we aim to connect the key advancements in RS using Generative Models (Gen-RecSys), encompassing: a foundational overview of interaction-driven generative models; the application of large language models (LLM) for generative recommendation, retrieval, and conversational recommendation; and the integration of multimodal models for processing and generating image and video content in RS. Our holistic perspective allows us to highlight necessary paradigms for evaluating the impact and harm of Gen-RecSys and identify open challenges.

Programs

Part 1: Introduction by Dr. Yong Zheng (20 mins)

Overview of Language Models and RSs

Overview of Language Modeling Paradigm in RSs

Part 2: Training Strategies of LLM-based RSs by Dr. Lemei Zhang (45 mins)

Pre-train, fine-tune paradigm for RSs

Prompting paradigm for RSs

Part 3: Optimization Objectives of LLM-based RSs by Dr. Peng Liu (20 mins)

Language modeling objectives to recommendation

Adaptive objectives to recommendation

Part 4: Ethical Issues and Trustworthiness of LLM-based RSs by Dr. Yashar Deldjoo (50 mins)

Different harm types, stakeholders involved, and harm severity in LLM-based RSs

Possible approaches to assessing and mitigating ethical issues and harms

Part 5: Evaluation and Available Resources by Dr. Peng Liu (20 mins)

Evaluation on recommendation accuracy and language perspectives

Open-sourced datasets and training platforms

Part 6: Summary and Future Directions by Dr. Jon Atle Gulla (20 mins)

Presenters

Yashar Deldjoo

Tenure-Track Assistant Professor

Polytechnic University of Bari
Italy

Zhankui He

Research Scientist

Google DeepMind
USA

Julian McAuley

Assistant Professor

UC San Diego
USA

Anton Korikov

PhD Student

University of Toronto
Canada

Scott Sanner

Associate Professor

University of Toronto
Canada

Arnau Ramisa

Research Scientist

Amazon
USA

Rene Vidal

>Research Scientist

Amazon and University of Pensilvania
USA

Maheswaran Sathiamoorthy

>Research Scientist

Bespoke Labs
USA

Atoosa Kasirzadeh

>Assistant Professor

University of Edinburgh
UK

Silvia Milano

>Assistant Professor

University of Exeter and LMU Munich
Germany

Materials

Slides: We are still working on it, and will upload soon!

Supplementary Materials:

Peng Liu, Lemei Zhang, and Jon Atle Gulla. "Pre-train, Prompt, and Recommendation: A Comprehensive Survey of Language Modeling Paradigm Adaptations in Recommender Systems", Transactions of the Association for Computational Linguistics 11 (2023): 1553-1571.
Yashar Deldjoo, Dietmar Jannach, Alejandro Bellogin, Alessandro Difonzo, and Dario Zanzonelli. "Fairness in recommender systems: research landscape and future directions", User Modeling and User-Adapted Interaction 34, no. 1 (2024): 59-108.