brycewang-stanford/orkg-api

Open Research Knowledge Graph (ORKG) API

Overview

The Open Research Knowledge Graph (ORKG) transforms unstructured scholarly articles into structured, machine-readable research contributions. Unlike traditional databases that store metadata (title, authors, DOI), ORKG captures the semantic content — research problems, methods, results, and their relationships. The REST API enables querying, creating, and comparing research contributions programmatically. Free, no authentication required for read operations.

API Endpoints

Base URL

https://orkg.org/api/

Search Papers

# Search papers in ORKG
curl "https://orkg.org/api/papers?q=climate+change+adaptation&size=20"

# Get paper details by ID
curl "https://orkg.org/api/papers/R12345"

Search Resources

# Search any resource (papers, predicates, comparisons)
curl "https://orkg.org/api/resources?q=machine+learning&size=20"

# Filter by class
curl "https://orkg.org/api/resources?q=BERT&exact=false&classes=Paper"

Comparisons

ORKG's unique feature — structured side-by-side comparison of papers:

# List comparisons
curl "https://orkg.org/api/comparisons?size=10"

# Get a specific comparison
curl "https://orkg.org/api/comparisons/R54321"

# Search comparisons
curl "https://orkg.org/api/comparisons?q=sentiment+analysis"

Research Contributions

# Get contributions of a paper
curl "https://orkg.org/api/papers/R12345/contributions"

# A contribution describes what a paper contributes:
# - Research problem addressed
# - Method used
# - Results achieved
# - Materials/datasets used

Python Usage

import requests

BASE_URL = "https://orkg.org/api"

def search_orkg_papers(query: str, size: int = 20) -> list:
    """Search papers in the Open Research Knowledge Graph."""
    resp = requests.get(f"{BASE_URL}/papers", params={"q": query, "size": size})
    resp.raise_for_status()
    data = resp.json()

    papers = []
    for item in data.get("content", []):
        papers.append({
            "id": item.get("id"),
            "title": item.get("title"),
            "created": item.get("created_at"),
            "contributions": item.get("contributions", [])
        })
    return papers

def get_paper_contributions(paper_id: str) -> dict:
    """Get structured research contributions for a paper."""
    resp = requests.get(f"{BASE_URL}/papers/{paper_id}/contributions")
    resp.raise_for_status()
    return resp.json()

def search_comparisons(topic: str) -> list:
    """Find structured paper comparisons on a topic."""
    resp = requests.get(f"{BASE_URL}/comparisons", params={"q": topic, "size": 10})
    resp.raise_for_status()
    return resp.json().get("content", [])

# Example usage
papers = search_orkg_papers("transfer learning NLP")
for p in papers:
    print(f"[{p['id']}] {p['title']}")

comparisons = search_comparisons("named entity recognition")
for c in comparisons:
    print(f"Comparison: {c.get('title')} ({len(c.get('contributions', []))} papers)")

Key Concepts

| Concept | Description | Example | |---------|-------------|---------| | Paper | A scholarly article with metadata | "Attention Is All You Need" | | Contribution | What a paper contributes to knowledge | "Proposes self-attention mechanism" | | Research Problem | The problem a contribution addresses | "Machine translation quality" | | Predicate | A relationship type | "has_method", "has_result", "uses_dataset" | | Comparison | Side-by-side structured comparison | "Transformer variants comparison" | | Resource | Any entity in the knowledge graph | A method, dataset, metric, or concept |

ORKG vs Traditional Databases

| Feature | Traditional (S2, Crossref) | ORKG | |---------|---------------------------|------| | Content | Metadata (title, DOI, citations) | Semantic content (methods, results) | | Structure | Flat records | Knowledge graph with relationships | | Comparison | Manual (read each paper) | Automated structured comparisons | | Machine-readable | Bibliographic metadata only | Research contributions structured | | Coverage | Broad (200M+ papers) | Deep but narrower (~50K papers) |

Use Cases

1. Literature surveys: Find existing comparisons to quickly understand a field
2. Method selection: Compare methods across papers on structured criteria
3. Gap analysis: Identify research problems without solutions
4. Reproducibility: Access structured descriptions of experimental setups

References

• ORKG Website
• ORKG API Documentation
• ORKG Help Center
• Jaradeh, M.Y., et al. (2019). "Open Research Knowledge Graph: Next Generation Infrastructure for Semantic Scholarly Knowledge." K-CAP 2019.