Cluster-Aware Dual-Level Test Specification Generation for Large-Scale Automotive Software Requirements

The paper introduces a "Cluster-then-Summarize" pipeline for generating test specifications that meet Automotive SPICE SWE.6 requirements, addressing the challenges of scaling to thousands of requirements. It employs sentence transformers for embedding, UMAP for dimensionality reduction, and HDBSCAN for clustering, with an automatic minimum cluster size selection based on quality metrics. The approach enhances integration test coverage and maintains summarization fidelity, leveraging a nearby-cluster context mechanism and Retrieval-Augmented Generation to align outputs with ISO 26262 and ASPICE standards, making it a significant advancement for automating test specification in large-scale automotive software development.