MatKG csv Dataset

This is the dataset for MatKG, the largest Knowledge Graph in the field of Material Science that is extracted from Scientific Literature. The dataset identifies some 200,000 concepts and 6.5 million relations between materials, their properties, applications, characterization methods, descriptors, synthesis methods, and symmetry labels expressed as a graph database. 

The subject_REL_object file contains the main list of entities and their relationships expressed as triple in the form [subject, rel, obj] with a 'count' paramater that is a measure of the relevance in the relation. 

The rawEntity_NER_DOI dataset gives the raw entities as extracted from the text along with their NER Tag and the DOI from which it was extracted. 

The rawEntity_Clean1_Clean2_Clean3_file gives the different cleaned version of the initially extracted raw text. 

An initial preprint of the data is available here: https://arxiv.org/abs/2210.17340