Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Scalable Construction of a Lung Cancer Knowledge Base: Profiling Semantic Reasoning in LLMs

Cesar Felipe Martínez Cisneros, Jesús Ulises Quiroz Bautista, Claudia Anahí Guzmán Solano, Bogdan Kaleb García Rivera, Iván García Pacheco, Yalbi Itzel Balderas Martínez, Kolawole John Adebayoc, Ignacio Arroyo Fernández

TL;DR

The paper tackles scalable construction of a domain-specific lung cancer knowledge base to support semantic reasoning in large language models. It introduces a two-phase OpenIE-based pipeline that filters PubMed content using MeSH terminology and CC0 licensing, followed by NER enrichment to produce a curated set of triplets for fine-tuning T5 models. Semantically supervised fine-tuning yields large gains on ROUGE metrics and moderate gains on BertScore, with strong convergence between model variants, suggesting that domain-specific pretraining can be mitigated by targeted semantic supervision. The work demonstrates the practicality of OpenIE-derived knowledge resources as a low-cost approach to enhance biomedical reasoning and points to future expansion to other diseases and improvements in triplet quality and factuality.

Abstract

The integration of Large Language Models (LLMs) into biomedical research offers new opportunities for domainspecific reasoning and knowledge representation. However, their performance depends heavily on the semantic quality of training data. In oncology, where precision and interpretability are vital, scalable methods for constructing structured knowledge bases are essential for effective fine-tuning. This study presents a pipeline for developing a lung cancer knowledge base using Open Information Extraction (OpenIE). The process includes: (1) identifying medical concepts with the MeSH thesaurus; (2) filtering open-access PubMed literature with permissive licenses (CC0); (3) extracting (subject, relation, object) triplets using OpenIE method; and (4) enriching triplet sets with Named Entity Recognition (NER) to ensure biomedical relevance. The resulting triplet sets provide a domain-specific, large-scale, and noise-aware resource for fine-tuning LLMs. We evaluated T5 models finetuned on this dataset through Supervised Semantic Fine-Tuning. Comparative assessments with ROUGE and BERTScore show significantly improved performance and semantic coherence, demonstrating the potential of OpenIE-derived resources as scalable, low-cost solutions for enhancing biomedical NLP.

Scalable Construction of a Lung Cancer Knowledge Base: Profiling Semantic Reasoning in LLMs

TL;DR

The paper tackles scalable construction of a domain-specific lung cancer knowledge base to support semantic reasoning in large language models. It introduces a two-phase OpenIE-based pipeline that filters PubMed content using MeSH terminology and CC0 licensing, followed by NER enrichment to produce a curated set of triplets for fine-tuning T5 models. Semantically supervised fine-tuning yields large gains on ROUGE metrics and moderate gains on BertScore, with strong convergence between model variants, suggesting that domain-specific pretraining can be mitigated by targeted semantic supervision. The work demonstrates the practicality of OpenIE-derived knowledge resources as a low-cost approach to enhance biomedical reasoning and points to future expansion to other diseases and improvements in triplet quality and factuality.

Abstract

The integration of Large Language Models (LLMs) into biomedical research offers new opportunities for domainspecific reasoning and knowledge representation. However, their performance depends heavily on the semantic quality of training data. In oncology, where precision and interpretability are vital, scalable methods for constructing structured knowledge bases are essential for effective fine-tuning. This study presents a pipeline for developing a lung cancer knowledge base using Open Information Extraction (OpenIE). The process includes: (1) identifying medical concepts with the MeSH thesaurus; (2) filtering open-access PubMed literature with permissive licenses (CC0); (3) extracting (subject, relation, object) triplets using OpenIE method; and (4) enriching triplet sets with Named Entity Recognition (NER) to ensure biomedical relevance. The resulting triplet sets provide a domain-specific, large-scale, and noise-aware resource for fine-tuning LLMs. We evaluated T5 models finetuned on this dataset through Supervised Semantic Fine-Tuning. Comparative assessments with ROUGE and BERTScore show significantly improved performance and semantic coherence, demonstrating the potential of OpenIE-derived resources as scalable, low-cost solutions for enhancing biomedical NLP.
Paper Structure (11 sections, 2 figures, 5 tables)

This paper contains 11 sections, 2 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Related Literature
  3. Theoretical Background
  4. Methodology and Implementation
  5. Methodology
  6. Results and Discussion
  7. Differential Metric Sensitivity to Semantic Fine-tuning
  8. Model Convergence Following Domain-Specific Fine-tuning
  9. Precision-Recall Rebalancing in ROUGE Metrics
  10. Conclusions
  11. Acknowledgments

Figures (2)

  • Figure 1: T5-Pubmed (Pre-finetuning)
  • Figure 2: T5-Pubmed (Post-finetuning)