Back to Blog
AI/ML

BioAgents: AI-Powered Knowledge Extraction for Research Papers

Discover how our AI agents automatically extract metadata, identify entities, and build knowledge graphs from research documents.

Ankita Choudhary
November 28, 2025
7 min read

Introducing BioAgents

BioAgents is our AI-powered system for automated knowledge extraction from research documents. Built on the Eliza OS agent framework, it transforms unstructured research papers into structured, queryable knowledge.

What BioAgents Can Do

1. Metadata Extraction

BioAgents automatically extracts:

  • Bibliographic data: Authors, titles, abstracts, keywords
  • Institutional affiliations: Universities, research centers, funding bodies
  • Publication details: DOIs, journal information, publication dates

2. Named Entity Recognition

Our specialized models identify:

  • Biological entities: Genes, proteins, diseases, compounds
  • Research concepts: Methods, techniques, experimental conditions
  • Geographic and temporal references: Study locations, time periods

3. Relationship Extraction

BioAgents identifies connections between entities:

  • Gene disease associations
  • Drug target interactions
  • Author collaboration networks
  • Citation relationships

How It Works

Document inputPDF, DOCXText extractionGROBID, Apache TikaSection classificationAbstract, methods, resultsEntity recognitionBioBERT, SciBERTRelationship extractionGraph neural networksKnowledge graphRDF / JSON-LD
From document to knowledge graph

Using BioAgents

typescript
// Processing is asynchronous - the call returns a task handle.
const res = await fetch('https://api.ekayana.com/api/bioagents/process', {
  method: 'POST',
  headers: {
    Authorization: `Bearer ${token}`,
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({
    cid: 'QmXg9Pp2ytZ14xgK35M6iTC2Vz6jR9zYgooNp2UHPTMnPN',
  }),
});

const { task_id } = await res.json();
// Poll GET /api/bioagents/status/{task_id} until it reports completion,
// then read the extracted entities from /api/bioagents/metadata.

// Query the resulting graph over SPARQL
const graph = await fetch('https://api.ekayana.com/api/kg/sparql', {
  method: 'POST',
  headers: {
    Authorization: `Bearer ${token}`,
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({
    query: `
    SELECT ?gene ?disease WHERE {
      ?gene bio:associatedWith ?disease .
      ?disease rdfs:label "breast cancer" .
    }
  `
});

Integration with DIDs

Every extracted knowledge graph is linked to the source document's DID, creating a verifiable chain of provenance:

json
{
  "@context": "https://schema.org",
  "@type": "Dataset",
  "identifier": "did:bio:devnet:9A3ck7prE8ZEAvQWt4jM2sEJJk7iWZZwoP7nnYoskcRA",
  "derivedFrom": "did:bio:devnet:3x3wS8ym8Z4MuANdeZ6CvpjFtvo6rF2NgGw92Qq5CLkw",
  "dateCreated": "2025-11-28",
  "creator": "bioagents:v2.1"
}

Learn more in our BioAgents Architecture documentation.

Ready to Get Started?

Explore our documentation to learn how to integrate Ekayana into your research workflow.

Read DocumentationMore Articles