A Tool for Automatically Cataloguing and Selecting Pre-Trained Models and Datasets for Software Engineering
Alexandra González, Oscar Cerezo, Xavier Franch, Silverio Martínez-Fernández
TL;DR
MLAssetSelection addresses the difficulty SE practitioners face in discovering suitable ML assets amid rapid proliferation of registries. It deploys a provider-agnostic three-layer web application (Angular frontend, FastAPI backend, PostgreSQL database) that continuously ingests assets, maps them to a taxonomy of 147 SE tasks, and presents SE-focused discovery through a unified Leaderboard and multi-criteria filters. A daily ingestion pipeline, a metrics-refresh job, and a Leaderboard Engine standardize and refresh evaluations, while an authenticated User Workspace enables personalized asset management and alerts, with export formats $CSV$, $JSON$, and $XML$ to support reproducibility. The ingestion pipeline was validated against expert judgment and a large language model, achieving almost perfect agreement with Cohen's kappa $k>0.8$, demonstrating reliable SE task mapping and evaluation standardization. Collectively, the work offers a practical, up-to-date tool that improves discoverability and selection of SE assets and outlines clear future directions for broader provider coverage and personalized recommendations.
Abstract
The rapid growth of machine learning assets has made it increasingly difficult for software engineers to identify models and datasets that match their specific needs. Browsing large registries, such as Hugging Face, is time-consuming, error-prone, and rarely tailored to Software Engineering (SE) tasks. We present MLAssetSelection, a web application that automatically extracts SE assets and supports four key functionalities: (i) a configurable leaderboard for ranking models across multiple benchmarks and metrics; (ii) requirements-based selection of models and datasets; (iii) real-time automated updates through scheduled jobs that keep asset information current; and (iv) user-centric features including login, personalized asset lists, and configurable alert notifications. A demonstration video is available at https://youtu.be/t6CJ6P9asV4.