WSDM 2024
136 papers
The 2024 edition of WSDM featured 209 papers, with a focus on addressing biases in recommendation systems and information retrieval. Research themes included correcting exposure and popularity bias through multifactorial causal debiasing and distributionally robust optimization. Considerable attention was given to differentiable ranking and sorting via continuous relaxations, as well as representation learning techniques designed to mitigate representation collapse and improve long-tail recommendations.
The conference also included work on sequential recommendation using linear recurrent units, adapter-based transfer learning, and multi-line mathematical formula recognition. Methodological developments involved meta-bootstrapping for link prediction and mixed censored multi-task learning for price prediction. Prolific authors contributing to the year's proceedings included Hao Peng, Philip S. Yu, Qishen Zhang, and Zhongyi Liu. The collected research highlights a trend toward improving the reliability and efficiency of high-dimensional learning models and unbiased learning to rank applications.
Top Research Directions
Research directions most represented by WSDM 2024 papers.
Top Authors
Impactful Papers
Papers from WSDM 2024 that shaped research clusters.
Influential
Pre-trained Recommender Systems: A Causal Debiasing Perspective
influential
Multimodal Feature Fusion for Recommende
Long-Term Value of Exploration: Measurements, Findings and Algorithms
influential
Mechanics and Remedies of Algorithmic Fi
Fly-Swat or Cannon? Cost-Effective Language Model Choice via Meta-Modeling
influential
Sequential Model Cascading for Cost Effi
K2: A Foundation Language Model for Geoscience Knowledge Understanding and Utilization
influential
Evaluating Large Language Models in Spec
Not All Negatives Are Worth Attending to: Meta-Bootstrapping Negative Sampling Framework for Link Pr
influential
Topology Constrained Sampling for Graph
SSLRec: A Self-Supervised Learning Framework for Recommendation
influential
Unifying Self-Supervised Learning via Sp
Long-Term Value of Exploration: Measurements, Findings and Algorithms
influential
Gradient Based Exploration in Neural Con
Dynamic Sparse Learning: A Novel Paradigm for Efficient Recommendation
influential
Non Uniform Embedding Dimension Allocati
Dynamic Sparse Learning: A Novel Paradigm for Efficient Recommendation
influential
The Hardware Software Mismatch in Sparse
Papers
« prev Page 3/3