HerbKG

HerbKG constructs a knowledge graph linking herbs, chemical constituents, genes, and diseases by extracting relations from over 500,000 PubMed abstracts to enable molecular interpretation of herbal medicine.

Key Features:

• Integration of herbal and molecular domains: Connects herbs to their chemical constituents, the genes affected by those chemicals, and associated diseases to represent cross-domain relationships.
• Automated construction framework: Uses an advanced learning framework to automate extraction and relation identification for scalable knowledge graph construction.
• Literature-scale extraction: Systematically organizes domain-specific findings derived from analysis of over 500,000 PubMed abstracts.
• Extensive knowledge repository: Populates the graph with over 53,000 relations extracted from the literature.

Scientific Applications:

• Drug discovery and development: Supports identification of herbal compounds and their molecular targets as candidate therapeutic agents.
• Personalized medicine: Enables exploration of gene–herb interactions to inform genotype-informed treatment hypotheses.
• Safety and efficacy assessments: Provides structured evidence to evaluate potential safety concerns and therapeutic efficacy of herbal treatments.

Methodology:

Automated extraction and relation identification from PubMed abstracts using an advanced learning framework to construct the knowledge graph and populate >53,000 relations.