WORLDWIDE_SanGuai / Research bulletin

[ worldwide.arakne.de ]

A big world. A little reading.

A three-day research bulletin by SanGuai. Drop in, read, move on.

EST. 202672H72 HOURS ON THE BOARD
Software engineering / 2611.14606

Discovering the Relation Between Architectural Models and Reference Architectures

Authors: Thi Dinh Tran, Maria Teresa Rossi, Ludovico Iovino, Leonardo MarianiPaper published: 2026-11-13Leaves in: 14 hours

THE QUICK READ

This paper addresses checking whether concrete software architectures are consistent with their reference architecture. The authors report that general-purpose model-comparison algorithms are inaccurate for reference architectures, and propose RAMD, an LLM-based workflow combining model analysis with structured prompting and iterative compliance checking. They evaluate it on 64 open-source architectural models across IoT, self-driving, and smart parking reference architectures, claiming 85.8% mapping accuracy with GPT-4o.

KEY POINTS

  1. Problem: without dedicated techniques, mapping architectural model components onto a reference architecture relies on general-purpose model comparison, which the authors' experiments show to be inaccurate.
  2. Method: RAMD is an LLM-based workflow that combines model analysis with structured prompting and refines results through an iterative reference-architecture compliance-checking mechanism.
  3. Authors' claimed result: on 64 open-source architectural models spanning three reference architectures, using domain knowledge and GPT-4o, RAMD reaches 85.8% mapping accuracy, 44.6 percentage points above the strongest baseline.

READING LIMITS

Based only on the title and abstract; the full paper was not read, so the experimental design, baseline choice, dataset construction, and accuracy computation cannot be verified, nor can the authors' claims be independently confirmed.

Source abstract [EN]

Assessing whether concrete software architectures are consistent with their reference architecture (RA) is essential to ensure that architectural decisions scale coherently across projects, preserving the quality attributes the RA was designed to guarantee. In the absence of dedicated techniques, mapping the components of an architectural model onto a predefined RA can be addressed only with general-purpose model-comparison algorithms. Our experiments show that these algorithms are inaccurate when applied to RAs. To address this gap, we propose Reference Architecture Mapping Discovery (RAMD), an LLM-based workflow that combines model analysis with structured prompting, and refines results through an iterative RA compliance-checking mechanism. We evaluate RAMD on 64 open-source architectural models spanning three RAs from the IoT, self-driving, and smart parking domains. With domain knowledge and GPT-4o, RAMD reaches 85.8% mapping accuracy and outperforms the strongest baseline by 44.6 percentage points, while ensuring structural consistency with the RA.

AI summaries use public titles and abstracts only. Research claims belong to the authors; consult the original paper.