Semantic Retrieval at Walmart, Text Embeddings in the Era of Large Language Models, and More!

Vol.82 for Dec 09 - Dec 15, 2024

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

This week’s newsletter highlights the following research:

1. An Efficient Two-Tower BERT Architecture for Large-Scale Product Search, from Walmart

2. A Unified Framework for Text Embeddings in the Age of Large Language Models, from Nie et al.

3. Arctic-Embed 2.0: Breaking the English-Multilingual Performance Trade-off in Text Embeddings, from Snowflake

4. JINA-CLIP-V2: A Three-Stage Framework for Cross-Modal and Text-Only Retrieval, from Jina AI

5. A Comprehensive Framework for Evaluating Document Parsing Systems, from Shanghai AI Laboratory

6. Amortized Inference for User History Modeling in Recommendation Systems, from LinkedIn

7. Leveraging Natural Language Preferences for Personalized Sequential Recommendations, from Meta

8. Exploring the Scaling Behavior of Transformer-based Sequential Recommendation Models, from Mercado Libre

9. Bridging the Gap Between Rerankers and Generators in Retrieval-Augmented Generation via Rationale Distillation, from CityU

10. Hierarchical Embedding Alignment Loss for Enhanced Retrieval and Representation Learning in RAG Systems, from Bhattarai et al.