The field of personalized recommendation systems is rapidly evolving, with a focus on developing innovative methods to capture dynamic user behaviors and preferences. Recent research has explored the use of generative models, such as diffusion models and transformer-based architectures, to improve the accuracy and efficiency of recommendation systems. Additionally, there is a growing interest in addressing challenges such as cold-start problems, data sparsity, and bias in user behavior logs. Noteworthy papers in this area include Spacetime-GR, which proposes a spacetime-aware generative model for large-scale online POI recommendation, and DiffusionGS, which introduces a novel approach to generative search with query-conditioned diffusion. These advancements have the potential to significantly enhance the performance and scalability of personalized recommendation systems.
Advances in Personalized Recommendation Systems
Sources
Similarity-Based Supervised User Session Segmentation Method for Behavior Logs
Spacetime-GR: A Spacetime-Aware Generative Model for Large Scale Online POI Recommendation
Modeling User Preferences as Distributions for Optimal Transport-based Cross-domain Recommendation under Non-overlapping Settings
Bootstrapping Conditional Retrieval for User-to-Item Recommendations
DiffusionGS: Generative Search with Query Conditioned Diffusion in Kuaishou
A Universal Framework for Offline Serendipity Evaluation in Recommender Systems via Large Language Models
Preference Trajectory Modeling via Flow Matching for Sequential Recommendation
HyST: LLM-Powered Hybrid Retrieval over Semi-Structured Tabular Data
DenseRec: Revisiting Dense Content Embeddings for Sequential Transformer-based Recommendation
Membership Inference Attacks on LLM-based Recommender Systems
Taming the One-Epoch Phenomenon in Online Recommendation System by Two-stage Contrastive ID Pre-training
Recycling History: Efficient Recommendations from Contextual Dueling Bandits
A Hybrid Recommendation Framework for Enhancing User Engagement in Local News
A Model-agnostic Strategy to Mitigate Embedding Degradation in Personalized Federated Recommendation
A Scenario-Oriented Survey of Federated Recommender Systems: Techniques, Challenges, and Future Directions
Anomaly Detection in Networked Bandits
ELIXIR: Efficient and LIghtweight model for eXplaIning Recommendations
MPFormer: Adaptive Framework for Industrial Multi-Task Personalized Sequential Retriever
Leveraging Generative Models for Real-Time Query-Driven Text Summarization in Large-Scale Web Search
SemSR: Semantics aware robust Session-based Recommendations
Addressing Personalized Bias for Unbiased Learning to Rank
OneRec-V2 Technical Report
Efficient Large-Scale Cross-Domain Sequential Recommendation with Dynamic State Representations