Abstract
The knowledge graph can link massive fragmented data, transform data into knowledge and provide services. However, there is a lack of related applications and research in the field of ocean engineering. In order to solve the above problems and fill the gap in the application research of related knowledge graphs in the field of ocean engineering, we design and implement knowledge graph based intelligent search system in ocean engineering. We define the ontology of the ocean engineering field and adopt a top-down approach to construct the knowledge graph. First, for processing unstructured data, we use BERT-BiLSTM-CRF for named entity recognition and use R-BERT for relationship extraction. Then, the knowledge graph is stored in the Neo4j graph database. Finally, the intelligent search system uses BERT-CRF and Lexicon Matching to parse the query and provide search services through Cypher sentence generation. In the entity extraction experiments of query parsing, we evaluated the BERT-CRF model performance for 100 to 575 project entities, and the experiments continue to improve as the number of project entities is increasing. Knowledge graph based intelligent search system in ocean engineering (OEIS) has been applied to Jiangsu’s comprehensive ocean management and monitoring. With the use of the system, we are able to collect more entity data, which will also effectively improve the performance of our intelligent search system.