RepoShapley: Shapley-Enhanced Context Filtering for Repository-Level Code Completion

2026-01-06

arXiv.org

AI Summary

This work addresses the challenge in warehouse-scale code completion where cross-file retrieved context often degrades generation quality due to conflicts or irrelevant content. To mitigate this, the authors propose a Shapley value–based context filtering framework that accurately evaluates and selects only the context truly beneficial for generation, leveraging offline annotations and a coalition-aware mechanism. The core innovation is the ChunkShapley module, which integrates teacher-forced likelihood, surrogate game modeling, exact Shapley value computation, and decoding-based optimal coalition validation, further enhanced by retrieval-augmented generation and bounded post-validation. Experiments demonstrate that the method significantly improves code completion quality across multiple benchmarks and backbone models while reducing the inclusion of harmful context and redundant retrieval overhead.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Mixture of Experts (MoE)
• Multiagent Systems: Mechanism Design

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Repository-level code completion benefits from retrieval-augmented generation (RAG). However, controlling cross-file evidence is difficult because chunk utility is often interaction-dependent: some snippets help only when paired with complementary context, while others harm decoding when they conflict. We propose RepoShapley, a coalition-aware context filtering framework supervised by Shapley-style marginal contributions. Our module ChunkShapley constructs offline labels by (i) single-chunk probing with teacher-forced likelihood to estimate signed, weighted effects, (ii) a surrogate game that captures saturation and interference, (iii) exact Shapley computation for small retrieval sets, and (iv) bounded post-verification that selects a decoding-optimal coalition using the frozen generator. We distill verified $KEEP$ or $DROP$ decisions and retrieval triggering into a single model via discrete control tokens. Experiments across benchmarks and backbones show that RepoShapley improves completion quality while reducing harmful context and unnecessary retrieval. Code: https://anonymous.4open.science/r/a7f3c9.

Problem

• repository-level code completion
• retrieval-augmented generation
• context filtering
• cross-file evidence
• harmful context

Innovation

• Shapley value
• repository-level code completion
• context filtering
• retrieval-augmented generation
• coalition-aware