G-WordNet: Moving WordNet 3.0 and its resources to a graph database

Abstract

In this paper, the convenience of storing a large lexical database (WordNet) in a graph database management system (Neo4j) is studied. The result is G-WordNet, which is a freely-available lexical database based on the Princeton WordNet 3.0 and all its sense-annotated corpora. We justify the need of this resource and the advantages of using a graph database in comparison with previous approaches. In addition, we present an application example of G-WordNet in the tasks of semantic lexical similarity using the declarative query language Cypher. Also, some possible usage scenarios of G-WordNet are discussed, particularly in the fields of lexicography, computational linguistics, natural language processing and NoSQL databases, for research, development, and teaching.