Can an LLM Detect Instances of Microservice Infrastructure Patterns?

Carlos Eduardo Duarte; Neil B. Harrison; Filipe Figueiredo Correia; Ademar Aguiar; Pavlína Gonçalves

cs.SEMar 24, 2026

Can an LLM Detect Instances of Microservice Infrastructure Patterns?

Authors: Carlos Eduardo Duarte, Neil B. Harrison, Filipe Figueiredo Correia, Ademar Aguiar, Pavlína Gonçalves

Organizations: ∗INESC TEC, Faculdade de Engenharia, Universidade do Porto, Porto, Portugal · †Department of Computer Science, Utah Valley University, Utah, USA · ‡INESC TEC, Porto, Portugal

Abstract

Architectural patterns are frequently found in various software artifacts. The wide variety of patterns and their implementations makes detection challenging with current tools, especially since they often only support detecting patterns in artifacts written in a single language. Large Language Models (LLMs), trained on a diverse range of software artifacts and knowledge, might overcome the limitations of existing approaches. However, their true effectiveness and the factors influencing their performance have not yet been thoroughly examined. To better understand this, we developed MicroPAD. This tool utilizes GPT 5 nano to identify architectural patterns in software artifacts written in any language, based on natural-language pattern descriptions. We used MicroPAD to evaluate an LLM's ability to detect instances of architectural patterns, particularly infrastructure-related microservice patterns. To accomplish this, we selected a set of GitHub repositories and contacted their top contributors to create a new, human-annotated dataset of 190 repositories containing microservice architectural patterns. The results show that MicroPAD was capable of detecting pattern instances across multiple languages and artifact types. The detection performance varied across patterns (F1 scores ranging from 0.09 to 0.70), specifically in relation to their prevalence and the distinctiveness of the artifacts through which they manifest. We also found that patterns associated with recognizable, dominant artifacts were detected more reliably. Whether these findings generalize to other LLMs and tools is a promising direction for future research.

Figures & tables

Work	Technique	Patterns	Language	Sample Size	P	R	A	F1
Daoudi et al. [ 7 ]	Heuristics	MVW	Bytecode	100	0.86	0.91	0.90	0.88
Haitzer and Zdun [ 8 ]	Heuristics	Any	-	-	-	-	-	-
Daniel et al. [ 9 ]	Heuristics	Database Per Service , Single Service Per Host , API Composition , Asynchronous Message , Command-Query Responsibility Segregation	SQL	-	-	-	-	-
Milhem et al. [ 10 ]	ML	Broker , Observer/Publish-Subscribe , Layered Architecture , Pipes and Filters , Shared Repository	Java	2	-	-	-	-
Chekhaba et al. [ 11 ]	ML	MVW	Java	265	-	-	-	0.56-0.68
Komolov et al. [ 12 ]	ML	MVW	Java	5,973	0.83	0.83	0.83	0.83

TABLE I : A summary of existing tools for detecting architectural patterns.

Fig. 1 : An overview of MicroPAD’s main components and interactions.

Metric	Min	Max	Mean	Median	SD
Age (years)	0.7	15.2	4.8	3.8	3.7
Contributors	1.0	8.0	1.7	1.0	1.2
Size (MB)	0.1	95.7	22.9	13.8	25.4
Stars	10.0	78,162.0	3,132.0	352.0	8,223.0

TABLE II : Characterization of the 190 analyzed repositories.

Panel of Table 4 — TABLE III : Various pattern-related metrics, including: prevalence of each pattern in the repositories, according to the repository contributors (PV); the tool’s precision (P), recall (R), accuracy (A), and F1 score (F1), per pattern; and the maximum file dominance index (Max FDI) value, per pattern.

	Predicted Positive	Predicted Negative
Actual Positive	114 (TP)	234 (FN)
Actual Negative	116 (FP)	1,246 (TN)

TABLE IV : Confusion matrix describing MicroPAD’s performance.

Pattern / File	Count	FDI
Service deployment platform
Makefile	120	79.12
main.yml	44	29.01
tsconfig.json	40	26.37
Service instance per container
Dockerfile	89	54.59

TABLE V: Top three dominant files per pattern (FDI).

Software Engineering

Software engineering studies how to design, build, test, deploy, and maintain reliable software systems and the processes behind them.

See all 270 papers

From Textual Requirements to Microservice Architectures - A Comprehensive Evaluation of LLM-Based Design Synthesis
Jul 30, 2026Danyllo Albuquerque, José Renan, Guillermo Rodríguez +6LLM Prompting Microservice Architecture
VIRTUS/UFCG Federal University of Campina Grande (UFCG), Paraiba Brazil · Federal University of Campina Grande (UFCG), Paraiba Brazil · Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Universidad Nacional del Centro de la Provincia de Buenos Aires (UNICEN), Argentina +2
LLM Code Smells: A Taxonomy and Detection Approach
May 21, 2026Zacharie Chenail-Larcher, Brahim Mahmoudi, Naouel Moha +2Static Code Analysis
Ecole de technologie sup´erieure, Montr´eal, Qu´ebec, Canada. · Universit´e du Qu´ebec `a Montr´eal, Montr´eal, Qu´ebec, Canada.
LLM Serving in the Wild: An Empirical Study of Frameworks, Methods, and System Designs
Aug 4, 2026Forough Majidi, Mohammad Mehdi Morovati, Foutse Khomh +1LLM Serving LLM Inference Acceleration
SWAT Lab., Polytechnique Montr´eal, Montr´eal, Canada

Jul 30, 2026J/K move · Enter open · S save

From Textual Requirements to Microservice Architectures - A Comprehensive Evaluation of LLM-Based Design Synthesis

Danyllo Albuquerque, José Renan, Guillermo Rodríguez, Guillermo Rodríguez+5

VIRTUS/UFCG Federal University of Campina Grande (UFCG), Paraiba Brazil · Federal University of Campina Grande (UFCG), Paraiba Brazil · Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Universidad Nacional del Centro de la Provincia de Buenos Aires (UNICEN), Argentina+2

LLM Prompting Microservice Architecture Software Architecture Design

Open paper AskPDF arXiv

Microservice architectures have become dominant for modernizing monolithic systems, yet identifying appropriate services remains challenging and largely manual. Existing decomposition approaches are predominantly code-centric, limiting applicability in early design stages where only textual requirements are available. Despite advances in Large Language Models (LLMs), limited empirical evidence exists on their ability to synthesize complete microservice architectures from natural-language requirements, including service definitions and inter-service interactions. This study investigates whether an LLM can bridge requirements engineering and architectural design, generating architectures solely from textual requirements and evaluating structural agreement and perceived quality of results. We conduct a mixed-method study using OpenAI o3 under zero-shot (ZS) and few-shot (FS) prompting across two systems (Bookstore, PetClinic), one execution per system/condition. Architectures are evaluated through (i) comparison with reference architectures using precision, recall, and F1-score for service identification and communication recovery, and (ii) a blinded expert assessment of correctness, completeness, modularity, and plausibility, plus open feedback synthesis. OpenAI o3 identifies services with higher agreement under FS prompting (F1 = 0.79 for ZS versus = 0.97 for FS). Communication recovery is more challenging: ZS produces dense architectures with high recall but low precision (F1 = 0.61), while FS improves agreement, reaching F1 = 0.82 and reducing unsupported dependencies. Expert evaluation corroborates these results, with FS architectures perceived as more modular, coherent, and plausible than ZS outputs. OpenAI o3 shows potential for requirements-driven synthesis when guided by exemplar prompting. Results are model- and context-specific from two small systems, not model-independent proof.

Abstract

Software Engineering

Explore similar work

From Textual Requirements to Microservice Architectures - A Comprehensive Evaluation of LLM-Based Design Synthesis

LLM Code Smells: A Taxonomy and Detection Approach

LLM Serving in the Wild: An Empirical Study of Frameworks, Methods, and System Designs