Ontology Learning and Knowledge Graph Construction: A Comparison of Approaches and Their Impact on RAG Performance
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917069492060160 |
|---|---|
| author | da Cruz, Tiago Tavares, Bernardo Belo, Francisco |
| author_facet | da Cruz, Tiago Tavares, Bernardo Belo, Francisco |
| contents | Retrieval-Augmented Generation (RAG) systems combine Large Language Models (LLMs) with external knowledge, and their performance depends heavily on how that knowledge is represented. This study investigates how different Knowledge Graph (KG) construction strategies influence RAG performance. We compare a variety of approaches: standard vector-based RAG, GraphRAG, and retrieval over KGs built from ontologies derived either from relational databases or textual corpora. Results show that ontology-guided KGs incorporating chunk information achieve competitive performance with state-of-the-art frameworks, substantially outperforming vector retrieval baselines. Moreover, the findings reveal that ontology-guided KGs built from relational databases perform competitively to ones built with ontologies extracted from text, with the benefit of offering a dual advantage: they require a one-time-only ontology learning process, substantially reducing LLM usage costs; and avoid the complexity of ontology merging inherent to text-based approaches. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_05991 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Ontology Learning and Knowledge Graph Construction: A Comparison of Approaches and Their Impact on RAG Performance da Cruz, Tiago Tavares, Bernardo Belo, Francisco Information Retrieval Artificial Intelligence Retrieval-Augmented Generation (RAG) systems combine Large Language Models (LLMs) with external knowledge, and their performance depends heavily on how that knowledge is represented. This study investigates how different Knowledge Graph (KG) construction strategies influence RAG performance. We compare a variety of approaches: standard vector-based RAG, GraphRAG, and retrieval over KGs built from ontologies derived either from relational databases or textual corpora. Results show that ontology-guided KGs incorporating chunk information achieve competitive performance with state-of-the-art frameworks, substantially outperforming vector retrieval baselines. Moreover, the findings reveal that ontology-guided KGs built from relational databases perform competitively to ones built with ontologies extracted from text, with the benefit of offering a dual advantage: they require a one-time-only ontology learning process, substantially reducing LLM usage costs; and avoid the complexity of ontology merging inherent to text-based approaches. |
| title | Ontology Learning and Knowledge Graph Construction: A Comparison of Approaches and Their Impact on RAG Performance |
| topic | Information Retrieval Artificial Intelligence |
| url | https://arxiv.org/abs/2511.05991 |