Newest papers

From Outcomes to Components: A Counterfactual Evaluation Framework for LLM-Proposed Repairs

Hengyu Ye, Songhua Hu · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.26650 · cited by 0 works (OpenAlex) · open-access PDF

Spanish

Una acción de corrección de la tasa de destino seleccionado por un modelo de lenguaje grande (LLM) especifica qué registros deben cambiar y cómo cambiar sus valores. Una acción fallida todavía puede contener una selección útil de registros de destino o una propuesta de actualización útil, dependiendo de cómo se combinan las dos. Introducimos un marco de evaluación de contratiempos que separa los registros seleccionados (tarjetas) del cambio aplicado (una actualización). Cuatro ejecuciones combinan los objetivos propuestos y las actualizaciones de referencia para medir los efectos de corregir o retener cada componente y su interacción. Evaluamos el marco en dos tareas: asignación de tráfico, donde la demanda afectada debe ser identificada de las cuentas de carretera, y reparación de bases de datos, donde las líneas afectadas deben ser identificadas de los informes

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

Abstract

A repair action proposed by a large language model (LLM) specifies which records to change and how to change their values. A failed action may still contain a useful choice of records or a useful update, depending on how the two are paired. We introduce a counterfactual evaluation framework that separates selected records (targets) from the applied change (an update). Four executions combine proposed and reference targets and updates to measure the effects of correcting or retaining each component and their interaction. We evaluate the framework on two tasks: traffic assignment, where affected demand must be identified from road counts, and database repair, where affected rows must be identified from aggregate reports. An injected fault supplies a reference action that reverses it, and the recovery rate measures an action's loss reduction relative to that reference. For sparse-target actions in the traffic experiments, retaining the proposed update on reference targets achieves 79.4% of the reference loss reduction, compared with 43.7% from tuning update strength on the proposed targets. In the database experiments, update replacement worsens 17 of 108 actions, all selecting unaffected records, with sharply different frequencies across models. The value of a proposed update depends on the targets it is paired with. Missed targets hide useful updates, while excess targets can make a milder update better than the reference correction. In a component-reuse study, testing new target-update pairings from two proposals raises mean held-out recovery rate to 0.616, compared with 0.580 when selecting between the original proposals.

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

Topics: Counterfactual thinking, Set (abstract data type), Aggregate (composite), Action (physics), Computer science, Aggregate data

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