TARQ: Pre-Ranking Framework Evaluation Data for Recommendation Systems

A dataset associated with the paper 'Equip Pre-ranking with Target Attention by Residual Quantization'. It supports a novel pre-ranking framework designed for industrial recommendation systems, aiming to balance efficiency and effectiveness. The dataset was uploaded by user 'zyody' and was last updated on 2026-05-27.

Use Cases

• Benchmarking pre-ranking model efficiency based on the described residual quantization framework.
• Evaluating the effectiveness of Target Attention (TA) mechanisms in latency-critical stages.
• Developing hybrid ranking systems that bridge the gap between candidate generation and final ranking.

Strengths

• Dataset is directly linked to a published research paper, providing academic context.
• The description specifies a clear, novel application for industrial recommendation systems.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and dataset scale are unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Associated with the paper 'Equip Pre-ranking with Target Attention by Residual Quantization'.

Freshness

Last updated 2026-05-27 03:21:18; freshness should be verified.