The field of recommender systems and information retrieval is witnessing significant advancements with a focus on enhancing performance, efficiency, and personalization. Recent developments highlight the importance of integrating multiple tasks, such as search and recommendation, to improve overall user experience. Innovations in models like Generative Search and Recommendation (GenSAR) and the introduction of novel frameworks such as VALUE (Value-Aware Large language model for query rewriting) demonstrate the push towards more sophisticated and user-centric approaches. Furthermore, advancements in techniques like safe screening rules for group OWL models and the development of efficient multi-task learning methods via generalist recommenders (GRec) aim to tackle challenges related to computational costs and scalability. Noteworthy papers include 'Lightweight and Direct Document Relevance Optimization for Generative Information Retrieval' which proposes a direct document relevance optimization method to improve retrieval effectiveness, and 'Efficient Multi-Task Learning via Generalist Recommender' which introduces a scalable and efficient approach to multi-task learning for recommender systems.
Advancements in Recommender Systems and Information Retrieval
Sources
Safe Screening Rules for Group OWL Models
Towards Principled Learning for Re-ranking in Recommender Systems
Lightweight and Direct Document Relevance Optimization for Generative Information Retrieval
GRIT: Graph-based Recall Improvement for Task-oriented E-commerce Queries
Multi-Perspective Attention Mechanism for Bias-Aware Sequential Recommendation
Efficient Multi-Task Learning via Generalist Recommender
RARe: Raising Ad Revenue Framework with Context-Aware Reranking
VALUE: Value-Aware Large Language Model for Query Rewriting via Weighted Trie in Sponsored Search
IterQR: An Iterative Framework for LLM-based Query Rewrite in e-Commercial Search System
xMTF: A Formula-Free Model for Reinforcement-Learning-Based Multi-Task Fusion in Recommender Systems
Unified Generative Search and Recommendation
To Match or Not to Match: Revisiting Image Matching for Reliable Visual Place Recognition
Joint Group Profiling and Recommendation via Deep Neural Network-based Multi-Task Learning
Diversity-aware Dual-promotion Poisoning Attack on Sequential Recommendation
InteractRank: Personalized Web-Scale Search Pre-Ranking with Cross Interaction Features
Unifying Search and Recommendation: A Generative Paradigm Inspired by Information Theory