Reducing the impact of out of vocabulary words in the translation of natural language questions into SPARQL queries

TL;DR

This paper presents a method combining Named Entity Linking, Recognition, and Neural Machine Translation to improve the translation of natural language questions into SPARQL, especially handling Out Of Vocabulary words more effectively.

Contribution

It introduces a novel approach that integrates NER, NEL, and NMT to enhance OOV word handling in question-to-SPARQL translation.

Findings

Outperforms existing methods on benchmark datasets

More resilient to OOV words in large ontologies

Improves accuracy of natural language to SPARQL translation

Abstract

Accessing the large volumes of information available in public knowledge bases might be complicated for those users unfamiliar with the SPARQL query language. Automatic translation of questions posed in natural language in SPARQL has the potential of overcoming this problem. Existing systems based on neural-machine translation are very effective but easily fail in recognizing words that are Out Of the Vocabulary (OOV) of the training set. This is a serious issue while querying large ontologies. In this paper, we combine Named Entity Linking, Named Entity Recognition, and Neural Machine Translation to perform automatic translation of natural language questions into SPARQL queries. We demonstrate empirically that our approach is more effective and resilient to OOV words than existing approaches by running the experiments on Monument, QALD-9, and LC-QuAD v1, which are well-known datasets for Question Answering over DBpedia.