A Unified Framework for Document Reranking with LLMs, Unifying Search and Recommendation Through Generative Retrieval, and More!

Vol.99 for Apr 07 - Apr 13, 2025

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

This week’s newsletter highlights the following research:

1. A Unified Framework for Document Reranking with Large Language Models, from Liu et al.

2. Enhancing Retrieval-Augmented Generation Through Dynamic Document Clustering and Compression, from Tsinghua University

3. Unifying Search and Recommendation Through Generative Retrieval, from Shi et al.

4. Addressing Cold-Start Problems in CTR Prediction via Supervised Diffusion Modeling, from Zhu et al.

5. Enhancing Search Pre-Ranking through Efficient Cross-Interaction Features, from Pinterest

6. A Dual-LLM Approach for User-Aligned Recommendation Exploration, from Google

7. Multi-modal Embeddings for Enhanced Product Search and Recommendation, from DoorDash

8. An Iterative Self-improving Framework for LLM-based Query Rewrite in E-commerce Search, from Meituan

9. Inference-Time Scaling of RAG Systems via Multi-Criteria Reranking, from Microsoft

10. Efficient Test-Time Augmentation for Sequential Recommendation, from NEU