cs.CLNov 7, 2026

Cascading Circle Loss: Order-Preserving Contrastive Learning for Graded Legal Retrieval

Authors: Rayane Kimouche, Ahmed Touila

Organizations: Dilitrust, Paris, France

Abstract

This paper introduces Cascading Circle Loss (CasCL), a contrastive learning framework for legal retrieval tasks with graded relevance. CasCL combines an InfoNCE objective for positive-negative separation with a cascade of ranking boundaries that model ordered relevance levels between queries and legal clauses. The framework also includes a negative-aware self-paced optimization mechanism and a relevance gain weighting strategy to improve ranking quality. Experiments on legal passage retrieval benchmarks show that CasCL consistently outperforms existing methods, including InfoNCE, Circle Loss, MNRL and Margin MSE, across both binary retrieval and graded ranking metrics.

Figures & tables

Appendix figures & tables3 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. GLIER: Generative Legal Inference and Evidence Ranking for Legal Case Retrieval

    Apr 26, 2026Minghan Li, Tianrui Lv, Chao Zhang +1Legal IRLearning to Rank

  2. NOWJ@COLIEE 2026: Adaptive Pipelines for Legal Retrieval and Reasoning

    Jul 18, 2026Thuong-Hieu Ngo, Hoang-Trung Nguyen, Huu-Dong Nguyen +5Legal NLPLegal IR

  3. Re-Ranking Through an Attribution Lens for Citation Quality in Legal QA

    Jun 2, 2026Mohamed Hesham Elganayni, Selim SalehCross-Encoder RerankingFeature Attribution