SPARQL-LLM 在 TEXT2SPARQL 2026 挑战赛中夺冠
原标题：SPARQL-LLM in the TEXT2SPARQL 2026 Challenge
AI 摘要
该论文介绍了开源知识图谱问答系统 SPARQL-LLM 参加 TEXT2SPARQL 2026 挑战赛的情况。SPARQL-LLM 在八个参赛方案中总排名第一，准确率最高比第二名高出 59%，并支持英语、德语和西班牙语的多语言问答。系统采用基于检索的方法，在查询前组装上下文，使每个问题的 LLM 交互次数固定，避免了智能体方法效率低、难以收敛的问题。该系统已在 GitHub 开源，并部署于 Expasy 生物信息学知识图谱的生产环境。
正文节选
*1 1 1 SPARQL-LLM in the TEXT2SPARQL 2026 Challenge Abstract Querying distributed data represented as Knowledge Graphs (KGs) via SPARQL remains a difficult task, requiring fluency in query syntax, endpoint specifications, and schema details. Knowledge Graph Question Answering (KGQA) systems address this by automatically translating natural language into SPARQL, and recent advances in Large Language Models (LLMs) have made this translation increasingly feasible. In this paper, we describe SPARQL-