[공지사항]을 빙자한 안부와 근황 
Show more

Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

KROMA: Ontology Matching with Knowledge Retrieval and Large Language Models

Created by
  • Haebom

Author

Lam Nguyen, Erika Barcelos, Roger French, Yinghui Wu

Outline

KROMA is a novel OM framework that dynamically enriches the semantic context of ontology matching (OM) tasks with structural, lexical, and definitional knowledge by leveraging large-scale language models (LLMs) within a retrieval augmented generation (RAG) pipeline. It is designed to address the limited adaptability of existing OM systems, and integrates similarity-based concept matching and a lightweight ontology refinement step to eliminate candidate concepts and significantly reduce the communication overhead caused by LLM invocations to improve performance and efficiency. Experiments on several benchmark datasets demonstrate that integrating knowledge retrieval and context-rich LLMs significantly improves the ontology matching performance, outperforming existing OM systems and state-of-the-art LLM-based approaches, while maintaining a similar communication overhead. This study highlights the feasibility and benefits of the proposed optimization techniques (targeted knowledge retrieval, prompt enrichment, and ontology refinement) for large-scale ontology matching.

Takeaways, Limitations

•
Takeaways:
◦
We demonstrate that leveraging the LLM and RAG pipelines can significantly improve the accuracy and efficiency of ontology matching.
◦
We empirically verify the effectiveness of optimization techniques such as targeted knowledge retrieval, prompt enrichment, and ontology improvement.
◦
It achieves performance that outperforms existing OM systems and state-of-the-art LLM-based approaches.
◦
We present a practical solution for large-scale ontology matching.
•
Limitations:
◦
Since the results are for a specific LLM and benchmark dataset, further research is needed to determine generalizability.
◦
It depends on the performance of LLM, and the limitations of LLM may also affect the performance of KROMA.
◦
The effectiveness of the lightweight ontology improvement step may vary depending on the characteristics of the dataset.
◦
The cost of LLM calls and the cost of building and maintaining a knowledge base need to be taken into consideration.
👍
Made with Slashpage