LRASGen: LLM-based RESTful API Specification Generation — RCR Report

This Replicated Computational Results (RCR) Report accompanies our TOSEM paper “LRASGen: LLM-based RESTful API Specification Generation”. LRASGen is the first framework to use Large Language Models (LLMs) for generating OpenAPI Specifications (OASs) directly from RESTful API source code, without requiring compilation or a runtime environment. The framework is organized as a six-step pipeline that includes endpoint entry file identification, endpoint entry code extraction and cleaning, endpoint method identification, parameter and response identification, parameter constraint identification, and OAS generation. To mitigate LLM hallucination, LRASGen employs three safeguards: structured-output enforcement , cross-validation across three LLMs , and a code-anchor mechanism . Evaluated on 53 real-world RESTful APIs spanning five programming languages and ten frameworks, LRASGen generated accurate specifications (e.g., 99.4% of endpoint methods) and covered, on average, 43.54% more entities than developer-provided specifications, outperforming the state-of-the-art technique, Respector, on all four identification tasks. To support replication, we released the source code, the enhanced ground truth (GT*), the generated specifications, and the experiment scripts. We also packaged the artifact as a Docker image, which needs no local Python installation. We archived the artifacts on Zenodo with a permanent DOI.

Authors

Institutions

Publication Details

Journal
ACM Transactions on Software Engineering and Methodology
Published
2026-11-17
DOI
https://doi.org/10.1145/3862595
Primary Topic
Software Engineering Research
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

LRASGen: LLM-based RESTful API Specification Generation — RCR Report

Dave Towey, Rubing Huang, Rongcun Wang, Man Zhang et al.
2 links
ACM Transactions on Software Engineering and Methodology
Software Engineering Research
article

LRASGen: LLM-based RESTful API Specification Generation — RCR Report

Dave Towey, Rubing Huang, Rongcun Wang, Man Zhang, Sida Deng, Chenhui Cui
article en

Abstract

This Replicated Computational Results (RCR) Report accompanies our TOSEM paper “LRASGen: LLM-based RESTful API Specification Generation”. LRASGen is the first framework to use Large Language Models (LLMs) for generating OpenAPI Specifications (OASs) directly from RESTful API source code, without requiring compilation or a runtime environment. The framework is organized as a six-step pipeline that includes endpoint entry file identification, endpoint entry code extraction and cleaning, endpoint method identification, parameter and response identification, parameter constraint identification, and OAS generation. To mitigate LLM hallucination, LRASGen employs three safeguards: structured-output enforcement , cross-validation across three LLMs , and a code-anchor mechanism . Evaluated on 53 real-world RESTful APIs spanning five programming languages and ten frameworks, LRASGen generated accurate specifications (e.g., 99.4% of endpoint methods) and covered, on average, 43.54% more entities than developer-provided specifications, outperforming the state-of-the-art technique, Respector, on all four identification tasks. To support replication, we released the source code, the enhanced ground truth (GT*), the generated specifications, and the experiment scripts. We also packaged the artifact as a Docker image, which needs no local Python installation. We archived the artifacts on Zenodo with a permanent DOI.

ACM Transactions on Software Engineering and Methodology
Macau University of Science and Technology (MO), University of Nottingham Ningbo China (CN), China University of Mining and Technology (CN), Beihang University (CN)
Software Engineering Research
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon