Contents I. Introduction
In 2012, Google proposed the concept of knowledge graph [1], since then large-scale domain knowledge graphs have been applied in many fields [2]. These graphs are supported by strong and clear semantic relationships, but lack of universality. At present, there are several issues with large-scale domain dynamic knowledge graphs, such as incomplete acquisition of original data, low accuracy with knowledge extraction and knowledge fusion, as well as chaotic semantic relations between entities [6]. This paper explores constructing dynamic knowledge graph based on ontology modeling and Neo4j graph database. The ontology data model constructed based on protégé and the “seven-step method” effectively avoids the filling of instances without concept classes in the original data, while removing concepts with low user attention or learning value, which ensures integrity of the original data acquisition, efficiency and accuracy of knowledge extraction and fusion, as well as rationality of the logical relations between classes. However, if we only consider the constraints of the ontology data model, the resulting knowledge graph will lack dynamic interactivity [7]. Considering the storage and query of complex relations, Neo4j graph database [8] is introduced and thus allowing a mapping between the ontology model and Neo4j. This method is applied in the field of agricultural informatization, and achieves dynamic domain knowledge graph with satisfying interactivity.
