In recent years, the field of neural machine translation (NMT) for SPARQL
query generation has witnessed a significant growth. Recently, the
incorporation of the copy mechanism with traditional encoder-decoder
architectures and the use of pre-trained encoder-decoders have set new
performance benchmarks. This paper presents a large variety of experiments that
replicate and expand upon recent NMT-based SPARQL generation studies, comparing
pre-trained and non-pre-trained models, question annotation formats, and the
use of a copy mechanism for non-pre-trained and pre-trained models. Our results
show that either adding the copy mechanism or using a question annotation
improves performances for nonpre-trained models and for pre-trained models,
setting new baselines for three popular datasets.

本文介绍了关于 NMT 的 SPARQL 查询语句生成的最新研究，并比较了预训练和非预训练模型、问题注释格式以及无需预训练和预训练模型的复制机制的使用。实验结果表明，无论是添加复制机制还是使用问题注释都可以提高非预训练模型和预训练模型的性能，为三个流行数据集设定新的基准线。