Exploring RAG's Vulnerability to Real-World Query Variations, Breaking the Bounded-Recall Barrier in Multi-Stage Ranking, and More!

Vol.100 for Apr 14 - Apr 20, 2025

Stay Ahead of the Curve with the Latest Advancements and Discoveries in Information Retrieval.

This week’s newsletter highlights the following research:

1. Exploring RAG's Vulnerability to Real-World Query Variations, from Cao et al.

2. Enabling Learned Sparse Retrieval with Large Language Models, from Amazon

3. Enhancing Knowledge Graph Construction with Document-Level Retrieval Augmentation, from Zhang et al.

4. A Bandit-Based Online Relevance Estimator for Efficient Document Ranking, from Rathee et al.

5. A Comprehensive Survey of Multimodal Retrieval-Augmented Generation, from Huawei

6. A Query Generation-Based, Training-Free Approach for LLM Recommendations, from KAIST

7. A Modular Approach to Retrieval-Augmented Generation with LLM Intrinsics, from IBM

8. A Framework for Managing Knowledge Conflicts in Retrieval-Augmented Generation, from Wang et al.

9. A Unified Approach to Uncertainty, Popularity and Exposure Bias in Session-based Recommender Systems, from University of Wroclaw

10. A Unified Visual Approach to Retrieval-Augmented Generation for Document Understanding, from Tanaka et al.