Construction of a GitHub Knowledge Graph based on DISTANER

This paper presents OpenKG, a GitHub Knowledge Graph constructed using the DISTANER framework, which combines distant supervision and active learning to achieve efficient named entity recognition (NER) in the open-source domain. OpenKG tackles the challenge of unstructured knowledge in GitHub repositories by extracting entities and relationships from semi-structured and unstructured data, storing them in a Neo4j database. The ontology design includes 13 entity types and 16 relationship types, capturing diverse aspects of OSS projects. DISTANER constructs an entity dictionary using custom lookup tables and external knowledge bases, enhancing NER performance while addressing annotation challenges. Evaluation results demonstrate the framework's effectiveness in accurately extracting entities and constructing a knowledge graph aligned with the ontology. OpenKG aims to enhance developer productivity and innovation by making open-source knowledge more accessible.