Geração de consultas SPARQL a partir de linguagem natural

2026-04-12 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Expert, quick

Summary

A new generic, domain-independent approach has been developed to generate SPARQL queries from natural language questions written in Portuguese. This method addresses the complexity of SPARQL, which typically limits access to Linked Open Data (LOD) repositories for non-expert users, and overcomes the English-centric and domain-specific limitations of existing solutions. The proposed system utilizes reference questions, parameterized query templates, and a synonym dictionary enhanced with lexical resources and similarity metrics. Implemented as the Natural2SPARQL tool, the approach was validated through a case study in the financial domain, using real data from the Brazilian stock exchange (B3). It achieved an end-to-end success rate of up to 93.3% in controlled settings, demonstrating flexibility, semantic accuracy, robustness, and low adaptation cost without requiring retraining, unlike learning-based methods.

Key takeaway

For NLP Engineers working with Portuguese-language data and Semantic Web technologies, this approach offers a robust, low-cost alternative to learning-based systems for SPARQL query generation. You should consider integrating template-based methods and enriched synonym dictionaries to improve accessibility to LOD repositories, especially in specialized domains like finance, without incurring retraining overheads.

Key insights

A domain-independent method generates SPARQL queries from Portuguese natural language using templates and a synonym dictionary.

Principles

• Query generation can be domain-independent.
• Lexical resources enhance synonym dictionaries.

Method

The method generates SPARQL queries by combining reference questions, parameterized query templates, and a synonym dictionary enriched with lexical resources and similarity metrics.

In practice

• Use Natural2SPARQL for Portuguese query generation.
• Apply template-based methods for low adaptation cost.

Topics

• SPARQL Query Generation
• Natural Language to SPARQL
• Portuguese Language Processing
• Domain-Independent NLP
• Natural2SPARQL Tool

Best for: AI Scientist, NLP Engineer, Research Scientist

Related on AIssential

• To Describe or Not to Describe? Benchmarking Database Representations for Schema Linking in Text-to-SQL — LLM-generated natural language descriptions enhance Text-to-SQL schema linking more effectively than DDL/DML.
• Solutions for Advanced NLP for Diverse Languages — spaCy's design and features support advanced NLP for diverse languages.
• Towards a Universal Dependencies Corpus for Portuguese Epidemiological Reports — Document extraction and representation significantly impact parsing epidemiological texts more than parser selection.
• Data Synthesis and Parameter-Efficient Fine-Tuning for Low-Resource NMT: A Case Study on Q'eqchi' Mayan — Synthetic data effectively bootstraps NMT structural learning for low-resource languages, but requires authentic data for semantic depth.
• Dissecting the MAD Landscape OS Space with DBpedia — Neural Machine Translation can effectively bridge the semantic gap between natural language and formal query languages like SPARQL.
• Telenor Nordics Customer Service self-help corpus — A new, ethically sourced, multilingual customer service self-help corpus for Nordic languages addresses a critical data scarcity for NLP research.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.