MatKG 1.4: Materials Science Knowledge Graph from 5 Million Papers

MatKG is a knowledge graph of materials science containing over 150,000 entities and 3.5 million unique triples. It was created by Vineeth Venugopal of MIT using natural language processing to extract entities and relationships from over 5 million scientific papers. The related publication is available in Nature Scientific Data.

Use Cases

• Semantic search for materials based on properties, applications, or synthesis methods mentioned in the description.
• Training link prediction models for materials science knowledge graphs based on the extracted entity relationships.
• Analyzing research trends by mining connections between materials, characterization methods, and applications from the literature corpus.

Strengths

• Contains over 150,000 entities and 3.5 million unique triples, making it the largest knowledge graph in its field at the time of writing.
• Built from a large corpus of over 5 million scientific papers using automated NLP methods.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

Massachusetts Institute of Technology

Collection Method

Automatically extracted using advanced natural language processing methods from scientific literature.