Home ➤ Projects ➤

Biomedical NLP

Biomedical texts contain vast and complex information, yet their richness often comes with challenges of heterogeneity, inconsistency, and semantic ambiguity. This project seeks to address these challenges by developing novel methods for biomedical natural language processing (NLP), with a particular focus on ontology-driven and embedding-based approaches.
We investigate how ontologies can enhance semantic representation and reasoning, enabling the detection of inconsistencies and improving interoperability across biomedical resources. Semantic search systems are designed to facilitate efficient access to knowledge within large biomedical datasets. Embedding-based approaches are explored to capture semantic relations beyond traditional ontology frameworks, enabling more flexible integration with neural NLP models.
The research also emphasizes building resources, benchmarks, and evaluation strategies that support the reproducibility and scalability of biomedical NLP. By integrating symbolic reasoning with modern embedding-based techniques, this work contributes toward more accurate, interpretable, and robust NLP systems for advancing biomedical knowledge discovery.

Objectives:

• Develop ontology-driven methods to improve semantic representation and reasoning in biomedical texts.
• Design semantic search and information extraction systems tailored for complex biomedical datasets.
• Investigate embedding-based approaches to capture nuanced semantic relations in biomedical language.
• Create frameworks to identify inconsistencies and errors in biomedical literature through ontology-based reasoning.
• Build interoperable resources, ontologies, and benchmarks to support biomedical NLP research and applications.

Keywords: Bioinformatics | Big Data | Natural Language Processing | Ontologies | Machine Learning / Deep Learning |

Publications

Team

External Collaborators: | Jingshan Huang | Fernando Gutierrez |