Fenced Citation-Context Retrieval for Case Law: Temporal Leakage and Degree Control Across Two Jurisdictions

2026-07-19

AI Summary

This work addresses temporal leakage in legal case retrieval caused by the use of future citation contexts, which inflates performance estimates. To mitigate this, the authors propose a temporal fencing mechanism that distinguishes genuine future citations from permissible pre-query references and indexing artifacts. They introduce a temporal admission decomposition method to quantify the proportion of illusory gains attributable to future evidence. Furthermore, they design a training-free anchored channel to enable fair and deployable evaluation. On the CLERC and ECtHR-PCR benchmarks, the fenced zero-training channel achieves a 16.1-point improvement in R@1000, with only 14.9% of the gain stemming from authentic future evidence. Notably, this zero-training approach attains an R@1000 of 79.56, rivaling the strongest trained systems.

Abstract

Incoming citation context-the text with which later cases describe a case when citing it-is a strong signal for legal precedent retrieval, but it is evaluated without a temporal fence: at query time the target precedent has not yet been cited by future cases, so an unfenced method credits itself with evidence unavailable at deployment. Our central contribution is to measure how much: a temporal-admission decomposition that splits the naive relax-the-fence gain into genuine future-citation leakage, legitimate pre-query admission, and an index effect, reporting the phantom fraction-the share that is future evidence. We instantiate this as a controlled deployability audit of incoming-citation retrieval across two jurisdictions, CLERC (US federal, 1.84M documents) and ECtHR-PCR (European Court of Human Rights), adding a zero-training anchor channel that admits only citers dated before the query. On ECtHR-PCR only 14.9% of the naive gain is genuine future evidence; on CLERC the over-credit grows to +4.95 R@1000 as the fence is relaxed. Under the fence, the zero-training channel gains +16.1 R@1000 over BM25 on CLERC, clears a citation-degree control, and reaches a published R@1000 estimate similar to the strongest trained system on ECtHR-PCR (79.56 vs 79.39) at zero training.

Problem

• temporal leakage
• legal precedent retrieval
• citation-context retrieval
• deployability audit
• future-citation

Innovation

• temporal leakage
• citation-context retrieval
• zero-training retrieval
• deployability audit
• legal precedent retrieval