The Critical Vulnerabilities of Dense Retrievers, A Survey of LLM Agents in Recommendation and Search, and More!

Vol.95 for Mar 10 - Mar 16, 2025

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

This week’s newsletter highlights the following research:

1. How Surface-Level Biases Undermine Dense Retrieval Models, from Fayyaz et al.

2. Teaching LLMs to Search On Demand Through Reinforcement Learning, from Renmin University

3. Gemini-Powered Embeddings for 250+ Languages, from Google DeepMind

4. A Survey of LLM Agents in Recommendation and Search, from Zhang et al.

5. Improving LLM-based Document Reranking via Reinforcement Learning-Enhanced Reasoning, from Zhuang et al.

6. A Comprehensive Framework for Evaluating LLMs Against Traditional Recommender Systems, from Liu et al.

7. Retrieval versus Generation: Leveraging Domain Knowledge at Inference Time for LLM Translation, from Google Translate

8. A Comprehensive Survey of Multi-Behavior Recommender Systems, from KAIST

9. Why PLM-Based Retrievers Favor LLM-Generated Content and How to Fix It, from Wang et al.

10. Enhancing RAG Systems Through Granularity-Aware Text Chunking, from Zhao et al.