Scientific Code Search at Scale: A Multi-Domain Dataset and Benchmark

DOI10.48550/arXiv.2607.05443MaRDI QIDQ7283958arXivFDO

Authors Ashkbiz Danehkar, Rahul Ramachandran, Nishan Pantha, Pranath Reddy Kumbam, Sajil Awale, Pushwitha Krishnappa, Muthukumaran Ramasubramanian, Nidhi Jha, Emily Foshee, Ankur Kumar, Rachel Slank

Publication date 3 July 2026

Published in arXiv

Tldr This paper presents a dataset and benchmark for scientific code search at scale, including a curated corpus of domain-classified repositories from NASA Science Mission Directorate divisions and novel information retrieval benchmarks designed by domain scientists. The two benchmarks evaluate repository search and large-scale code snippet retrieval across seven programming languages, revealing significant performance variation across scientific domains.

Full work available at URL https://doi.org/10.48550/arXiv.2607.05443

arXiv classification

cs.AI cs.IR cs.SE

This page was built for publication: Scientific Code Search at Scale: A Multi-Domain Dataset and Benchmark

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7283958)