一句话：提出 GADM 框架与领域自适应判定预言，从经济价值与合规性角度检测数据栈中的治理缺陷。
数据密集型软件工程已有成熟的吞吐与延迟基准（如 TPC-H），却缺乏针对业务语义正确性的判定预言（oracle），即数据资产是否具备经济效用（VRIN）与合规性。高效处理不合规或低价值数据的流水线实际存在“治理缺陷”，而传统测试套件无法察觉。本文提出 Governance-Aware Data Monetization（GADM）框架，一种用于评估数据资产就绪度的领域自适应判定预言。作者基于对 520 篇同行评议文献的系统综述，形式化了“数据缺陷”分类体系，涵盖策略违规、血缘断链与价值脱节，并给出六阶段持续验证流水线，作为按资源基础观（RBV）准则评估数据资产的质量门禁。评估通过真实案例的回顾性缺陷分析完成，包括 Netflix 的隐私失败以及西门子与 Mayo Clinic 的价值捕获，表明该框架能识别纯技术基准遗漏的缺陷。
While data-intensive software engineering has established robust benchmarks for throughput and latency (e.g., TPC-H), it lacks realistic oracles for business-semantic correctness—specifically, whether data assets possess economic utility (VRIN) and regulatory compliance. A data pipeline that efficiently processes non-compliant or low-value data effectively exhibits a "governance bug", yet traditional test suites fail to detect these defects. This paper presents the Governance-Aware Data Monetization (GADM) Framework, a domain-adaptive oracle designed to benchmark the readiness of data assets. Grounded in a Systematic Literature Review (SLR) of 520 peer-reviewed sources, we formalize a taxonomy of "data defects"-including policy violations, lineage gaps, and value disconnects. The GADM Oracle specifies a six-phase continuous verification pipeline, acting as a quality gate that evaluates data assets against Resource-Based View (RBV) criteria. We evaluate the Oracle's efficacy through retrospective bug analysis of real-world cases, including privacy failures at Netflix and value capture at Siemens and May Clinic, demonstrating how the framework identifies defects that purely technical benchmarks miss.
@inproceedings{WeerasingheAC26,
title = {GADM-Oracle: A Domain-Adaptive Oracle for Detecting "Data Bugs" in Enterprise Data Stacks},
author = {Shakthi Weerasinghe and Amr S. Abdelfattah and Tomas Cerny},
booktitle = {FSE},
year = {2026},
doi = {10.1145/3803437.3806093}
}