Related Experiment Video
Updated: Dec 23, 2025
05:47
Evidence-based Knowledge Synthesis and Hypothesis Validation: Navigating Biomedical Knowledge Bases via Explainable AI and Agentic Systems
Published on: June 13, 2025
1.1K
Your relevance feedback is essential: enhancing the learning to rank using the virtual feature based logistic
Fei Cai1, Deke Guo, Honghui Chen
1Science and Technology on Information Systems Engineering Laboratory, National University of Defense Technology, Changsha, China.
Plos One
|December 20, 2012
Summary
This study introduces a Virtual Feature based Logistic Regression (VFLR) model for information retrieval ranking. VFLR improves retrieval performance by incorporating user feedback through principal component analysis, outperforming existing methods.
Area of Science:
- Computer Science
- Information Science
Background:
- Information retrieval systems require effective ranked lists for output.
- Existing ranking models often lack user feedback integration, leading to high computation and low performance.
- Indefinite query expressions pose challenges for traditional retrieval methods.
Purpose of the Study:
- To propose a novel ranking method, Virtual Feature based Logistic Regression (VFLR), that leverages user relevance feedback.
- To address the limitations of existing methods by incorporating user interaction data.
- To enhance the accuracy and efficiency of ranked list generation in information retrieval.
Main Methods:
- Developed a Virtual Feature based Logistic Regression (VFLR) ranking approach.
- Extracted essential and independent virtual features (VF) using principal component analysis (PCA).
- Integrated user relevance feedback into the VF extraction process.
Main Results:
- VFLR demonstrated superior performance compared to state-of-the-art methods on LETOR 4.0 datasets.
- Achieved improvements in Mean Average Precision (MAP), Precision at k (P@k), and Normalized Discounted Cumulative Gain at k (NDCG@k).
- The method effectively utilizes user feedback for enhanced ranking.
Conclusions:
- The proposed VFLR method offers a significant advancement in learning effective ranking models.
- Incorporating user relevance feedback via virtual features enhances retrieval performance.
- VFLR provides a computationally efficient and accurate solution for information retrieval ranking problems.
