Querying the DepthsUnveiling the Strengths and Struggles of Large Language Models in SPARQL Generation
- Ghajari, Adrián
- Ros, Salvador
- Pérez, Álvaro
ISSN: 1135-5948
Ano de publicación: 2024
Número: 73
Páxinas: 271-281
Tipo: Artigo
Outras publicacións en: Procesamiento del lenguaje natural
Resumo
La irrupción de la Web Semántica ha precipitado una proliferación de datos estructurados manifestados en forma de grafos de conocimiento, subrayando la imperativa necesidad de interfaces de lenguaje natural para mejorar la accesibilidad a estos repositorios de información. La capacidad de articular consultas en lenguaje natural y posteriormente recuperar datos a través de consultas SPARQL asume una importancia primordial. En la presente investigación, hemos analizado la eficacia de la técnica de in-context learning usando una arquitectura basada en agentes para facilitar la construcción de consultas SPARQL. Contrariamente a las expectativas iniciales, se ha encontrado que la mejora del prompt de in-context learning a través de mecanismos basados en agentes disminuye la eficacia de los Sistemas Basados en Modelos de Lenguaje (LLMS), al ser percibidos como ”ruido” extrínseco, mostrando así las limitaciones inherentes de esta aproximación. Los resultados resaltan la necesidad de profundizar en las técnicas de entrenamiento y fine-tuning de modelos, centrándose en los aspectos relacionales de los esquemas de ontología.
Referencias bibliográficas
- Dorobat¸, I. C. and V. Posea. 2020. onIQ: An Ontology-Independent Natural Language Interface for Building SPARQL Queries. In 2020 IEEE 16th International Conference on Intelligent Computer Communication and Processing (ICCP), pages 139–144, September.
- Golovneva, O., Z. Allen-Zhu, J. Weston, and S. Sukhbaatar. 2024. Reverse Training to Nurse the Reversal Curse, March. arXiv:2403.13799 [cs].
- Guo, D., Q. Zhu, D. Yang, Z. Xie, K. Dong, W. Zhang, G. Chen, X. Bi, Y. Wu, Y. K. Li, F. Luo, Y. Xiong, and W. Liang. 2024. DeepSeek-Coder: When the Large Language Model Meets Programming – The Rise of Code Intelligence, January. arXiv:2401.14196 [cs].
- He, S., Y. Zhang, K. Liu, and J. Zhao. 2014. CASIA@V2: A MLN-based Question Answering System over Linked Data. In Working Notes for CLEF 2014 Conference, Sheffield, UK.
- Jiang, A. Q., A. Sablayrolles, A. Mensch, C. Bamford, D. S. Chaplot, D. d. l. Casas, F. Bressand, G. Lengyel, G. Lample, L. Saulnier, L. R. Lavaud, M.-A. Lachaux, P. Stock, T. L. Scao, T. Lavril, T. Wang, T. Lacroix, and W. E. Sayed. 2023. Mistral 7B, October. arXiv:2310.06825 [cs].
- Kaufmann, E., A. Bernstein, and R. Zumstein. 2006. Querix: A Natural Language Interface to Query Ontologies Based on Clarification Dialogs. In Proceedings of the 5th International Semantic Web Conference, Athens.
- Li, Y., H. Yang, and H. Jagadish. 2007. NaLIX: A Generic Natural Language Search Environment for XML Data. acmtds, accepted. ACM Trans. Database Syst., 32, November.
- Liang, S., K. Stockinger, T. M. de Farias, M. Anisimova, and M. Gil. 2021. Querying knowledge graphs in natural language. Journal of Big Data, 8(1):3, January.
- Perevalov, A., X. Yan, L. Kovriguina, L. Jiang, A. Both, and R. Usbeck. 2022. Knowledge Graph Question Answering Leaderboard: A Community Resource to Prevent a Replication Crisis, January. arXiv:2201.08174 [cs].
- Popescu, A.-M., O. Etzioni, and H. Kautz. 2003. Towards a theory of natural language interfaces to databases. In Proceedings of the 8th international conference on Intelligent user interfaces, IUI ’03, pages 149–157, New York, NY, USA. Association for Computing Machinery.
- Qi, P., Y. Zhang, Y. Zhang, J. Bolton, and C. D. Manning. 2020. Stanza: A Python Natural Language Processing Toolkit for Many Human Languages, April. arXiv:2003.07082 [cs].
- Rony, M. R. A. H., U. Kumar, R. Teucher, L. Kovriguina, and J. Lehmann. 2022. SGPT: A Generative Approach for SPARQL Query Generation From Natural Language Questions. IEEE Access, 10:70712–70723.
- Rozière, B., J. Gehring, F. Gloeckle, S. Sootla, I. Gat, X. E. Tan, Y. Adi, J. Liu, R. Sauvestre, T. Remez, J. Rapin, A. Kozhevnikov, I. Evtimov, J. Bitton, M. Bhatt, C. C. Ferrer, A. Grattafiori, W. Xiong, A. D´efossez, J. Copet, F. Azhar, H. Touvron, L. Martin, N. Usunier, T. Scialom, and G. Synnaeve. 2024. Code Llama: Open Foundation Models for Code, January. arXiv:2308.12950 [cs].
- Soru, T., E. Marx, A. Valdestilhas, D. Esteves, D. Moussallem, and G. Publio. 2018. Neural Machine Translation for Query Construction and Composition, July. arXiv:1806.10478 [cs].
- Taffa, T. A. and R. Usbeck. 2023. Leveraging LLMs in Scholarly Knowledge Graph Question Answering, November. arXiv:2311.09841 [cs].
- Trivedi, P., G. Maheshwari, M. Dubey, and J. Lehmann. 2017. LC-QuAD: A Corpus for Complex Question Answering over Knowledge Graphs. In The Semantic Web – ISWC 2017: 16th International Semantic Web Conference, Vienna, Austria, October 21-25, 2017, Proceedings, Part II, pages 210–218, Berlin, Heidelberg, October. Springer-Verlag.
- Wang, C., M. Xiong, Q. Zhou, and Y. Yu. 2007. PANTO: A Portable Natural Language Interface to Ontologies. In The Semantic Web: Research and Applications, volume 4519, pages 473–487. Book Title: The Semantic Web: Research and Applications ISSN: 0302-9743, 1611-3349 Series Title: Lecture Notes in Computer Science.
- Yang, S., M. Teng, X. Dong, and F. Bo. 2023. LLM-Based SPARQL Generation with Selected Schema from Large Scale Knowledge Base. In H. Wang, X. Han, M. Liu, G. Cheng, Y. Liu, and N. Zhang, editors, Knowledge Graph and Semantic Computing: Knowledge Graph Empowers Artificial General Intelligence, Communications in Computer and Information Science, pages 304–316, Singapore. Springer Nature.