EP-TransPE Benchmark Dataset
Published: 7 October 2026| Version 2 | DOI: 10.17632/5w5gv557s6.2
Contributor:
Description
This repository contains the benchmark metadata, evaluation records, human assessment data, post-editing experiment records, retrieval overlap audit files, and analysis scripts supporting: "A provenance-controlled retrieval-augmented large language model framework for technical translation in energy and power engineering" submitted to Computers & Electrical Engineering.
Files
Institutions
- Shenyang Institute of EngineeringLiaoning, Shenyang
Categories
Artificial Intelligence, Information Retrieval, Computational Linguistics, Natural Language Processing, Corpus-Based Translation Studies