Newest papers

Which Code Changes Need Re-embedding? CIPBench and the Limits of Multi-Signal Index Maintenance

Nannan Zhao, Jiameng Zhang, Taoyu Zhong, Hao Wang, Bo Wang · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.25108 · cited by 0 works (OpenAlex) · open-access PDF

Korean

벡터 인덱스를 지속적으로 진화하는 코드와 동기화하도록 유지해야하는 텍스트 생성 시스템이 녹음하기 전에 녹음하는 것이 아니지만 인덱스 구조가 실제로 새로운 삽입을 필요로 하는지 결정하는 원칙적 근거는 없습니다. 우리는 벡터 인덱스를 지속적으로 진화하는 코드와 동기화하기 전에 텍스트의 삽입 효과를 예측할 수 있는지 여부를 묻습니다. Python, Go, JavaScript에서 10 개의 오픈 소스 리포터에서, 우리는 변화가 단일 신호 교체를 통해 교체된 실질성으로부터 진화하는 방법을 특징화합니다. 링크는 정보적이지만 결정적인 것이 아닙니다: 합성 클래스는 교체를 급격하게 분리하지만, 영향은 단위의 노출에 달려 있으며

Machine translation (Meta M2M100); the English original follows.

Abstract

Repository-level retrieval-augmented generation systems must keep vector indexes synchronized with continuously evolving code, yet there is no principled basis for deciding which changed code units actually need new embeddings. We ask whether the retrieval impact of a code change can be anticipated before that embedding is computed. Across 10 open-source repositories in Python, Go, and JavaScript, we characterize how a change propagates from syntactic substantiality through embedding displacement to retrieval impact. The links are informative but not deterministic: syntactic class separates displacement sharply, yet impact depends on a unit's retrieval exposure as much as on how far its representation moves, and recurring counterexamples bound what any single signal can decide. We formalize the decision as CIPBench, a benchmark that scores maintenance policies under a strict pre-embedding information boundary. Prediction works, and one signal exhausts it. A ranker using only a unit's change history substantially reduces retrieval harm relative to random and recency-based scheduling. Combining it with three further signals--individually discriminative and largely unassociated with one another--leaves the decision statistically unchanged across every budget we evaluate, for reasons we trace to the structure of the score rather than to its tuning. What pays off is not ranking but filtering: units whose text is unchanged from the indexed state can be skipped with no risk and, in the most reversion-heavy project, at nearly half the embedding cost. Signal-level informativeness is therefore not evidence of decision-level value, and separating the two requires evaluating the decision a signal produces rather than its association with the outcome.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Discriminative model, Embedding, Computer science, Ranking (information retrieval), Benchmark (surveying), Counterexample

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange