Fast and Efficient Multi-Vector Search, A Parameter-Space Approach to Retrieval Augmented Generation, and More!

Vol.89 for Jan 27 - Feb 02, 2025

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

This week’s newsletter highlights the following research:

1. Accelerating Multi-Vector Retrieval through Optimized Architecture and Efficient Scoring, from Scheerer et al.

2. A Single Foundation Model for Large-Scale Personalized Recommendation, from LinkedIn

3. A Parameter-Space Approach to Retrieval Augmented Generation, from Tsinghua University

4. Chain-of-Thought Retrieval Augmented Generation for Complex Queries, from Microsoft

5. Time-Sensitive User Preference Learning in Streaming Recommendations, from ByteDance

6. Zero-Shot Document Re-ranking via Answer Scent for Enhanced Information Retrieval, from UIBK

7. Enhancing RAG Performance Through Collaborative Multi-Agent Learning, from Baidu

8. Matryoshka Architecture for Efficient Text Re-Ranking, from Liu et al.

9. Aligning LLM-based Retrieval with Data Organization for Complex Queries, from Chen et al.

10. Adaptive Hard Negative Mining for Zero-Shot Dense Retrieval, from the University of Amsterdam