ARUQULA - An LLM based Text2SPARQL Approach using ReAct and Knowledge Graph Exploration Utilities - based on SPINACH

Our contribution to the 1st Text2Sparql-Challenge, colocated with Text2KG at ESWC25. This code is based on SPINACH (Publication / Code).

Authors: Felix Brei* and Lorenz Bühmann* and Johannes Frey* and Daniel Gerber* and Lars-Peter Meyer* and Claus Stadler* and Kirill Bulert

* authors contributed equally

Challenge Result (published June 2025):

• 🥇 Best performance in the category "Corporate"
• 🥇 Best performance in the category "DBpedia english"
• 🥈 second in overal performance

All results: https://text2sparql.aksw.org/results/

Citation

Our contribution to the 1st Text2Sparql-Challenge is peer reviewed and published at Ceur-WS and as preprint at arxiv:

@InProceedings{Brei2025AruqulaLlmBasedText2Sparql,
  author    = {Felix Brei and Lorenz Bühmann and Johannes Frey and Daniel Gerber and Lars-Peter Meyer and Claus Stadler and Kirill Bulert},
  booktitle = {Proceedings of the First International TEXT2SPARQL Challenge, Co-Located with Text2KG at ESWC25, June 01, 2025, Portorož, Slovenia},
  title     = {{ARUQULA} - An {LLM} based {Text2SPARQL} Approach using {ReAct} and Knowledge Graph Exploration Utilities},
  year      = {2025},
  editor    = {Sebastian Tramp and Marcos Gôlo and Edgard Marx and Paulo Viviurka do Carmo},
  pages     = {49-63},
  series    = {{CEUR} Workshop Proceedings},
  volume    = {4094},
  doi       = {10.48550/arXiv.2510.02200},
  url       = {https://ceur-ws.org/Vol-4094/paper4.pdf},
}