Paper Digest: CIKM 2025 Papers & Highlights

2025-11-10 · Source: Information Retrieval – Resources | Paper Digest · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Cybersecurity & Data Privacy · Depth: Advanced, extended

Summary

This document provides a curated digest of 500 accepted papers from the ACM Conference on Information and Knowledge Management (CIKM) 2025, selected by the Paper Digest Team's daily digest algorithm. CIKM is an annual computer science research conference focusing on information and knowledge management. Each paper entry includes its title, authors, and a highlight sentence summarizing its main topic or proposed solution. The digest aims to help readers quickly grasp the core ideas of the presented research, covering diverse areas such as Large Language Models (LLMs) for recommendation, graph neural networks, time series forecasting, privacy-preserving techniques, and multimodal data analysis. The Paper Digest platform also offers services for searching, reviewing, and generating reports on these papers.

Key takeaway

AI Scientists developing advanced recommendation systems or robust AI models should explore the diverse methodologies presented at CIKM 2025. Focus on integrating LLMs with graph-based learning and multimodal data to tackle challenges like data sparsity, bias, and real-time performance. Consider adopting frameworks that offer explainability and privacy guarantees, as these are critical for deploying reliable and ethical AI solutions in industrial settings.

Key insights

CIKM 2025 research highlights advancements in LLMs, graph learning, and multimodal data, addressing challenges in recommendation, forecasting, and privacy.

Principles

• LLMs are increasingly integrated across diverse tasks, from recommendation to reasoning.
• Graph-based methods are crucial for modeling complex relationships and dynamic data.
• Addressing data sparsity, bias, and privacy remains central to advanced AI systems.

Method

Many papers propose novel frameworks and algorithms, often combining deep learning (e.g., Transformers, GNNs, Diffusion Models) with techniques like contrastive learning, reinforcement learning, and knowledge graphs to overcome specific data or model limitations.

In practice

• Utilize LLMs for query understanding and personalized content generation in various domains.
• Employ graph neural networks for complex data modeling in recommendation and anomaly detection.
• Implement privacy-preserving techniques like homomorphic encryption in federated learning.

Topics

• Large Language Models
• Recommendation Systems
• Graph Neural Networks
• Time Series Forecasting
• Federated Learning

Code references

• westlake-repl/microlens
• thu-keg/crab
• thu-keg/llmael
• idea-isail-lab-uiuc/climatebench-m
• idea-isail-lab-uiuc/pyg-ssl

Best for: AI Scientist, AI Researcher, Research Scientist, AI Student

Related on AIssential

• Guiding Federated Graph Recommendation with LLM-encoded knowledge — LLM-encoded semantic knowledge can guide federated graph recommendation, improving accuracy by selectively aggregating structural representations across clients.
• Most Influential IJCAI Papers (2026-03 Version) — Citation-based rankings reveal key trends and enduring impact in AI research from IJCAI.
• IEEE Transactions on Artificial Intelligence, Volume 7, Issue 2, February 2026 — The volume showcases diverse AI advancements, from LLM applications in autonomous driving to specialized models for protein language processing.
• 2025 Year in Review — Integrating LLMs with RecSys can yield powerful, steerable recommendation systems capable of cross-platform suggestions.
• A Production System for Podcast Discovery, A Fully Open-Source Frontier Search Agent, and More! — Recent research advances information retrieval through LLMs, sparse autoencoders, and multimodal embeddings.
• Are Large Language Models Suitable for Graph Computation? Progress and Prospects — LLMs show promise for simple graph tasks but struggle with large-scale, exact graph computation, necessitating role-based integration.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Information Retrieval – Resources | Paper Digest.