Construction Of Knowledge Graph To Diagnose For Aquatic Animals Disease

China is a large fishery country,and aquatic animal diseases have a serious adverse impact on aquaculture industry.In this paper,the joint extraction of entity relations for knowledge graph of aquatic animal disease diagnosis is investigated to provide support for aquatic animal disease diagnosis.Entity identification and relationship extraction are some of the main key techniques for constructing knowledge graphs for aquatic animal disease diagnosis.However,the data of aquatic animal diseases are widely distributed and very fragmented and disordered.The urgent problem is to organize and share these data resources in a rational way.Since the collected text corpus of aquatic animal diseases is chapter-level data,the difficulty in the task of entity and relationship extraction lies in the large span between two entities and the existence of a large number of overlapping relationships,etc.These problems will affect the accuracy of information extraction and thus the quality of building knowledge graphs for aquatic animal disease diagnosis.To address the above problems,the main research work of this paper is as follows.(1)A domain corpus of aquatic animal diseases was constructed.In this paper,we proposed the H-BIO annotation method and selected the annotation tool UltraEdit to annotate the pre-processed Chinese aquatic animal disease corpus,and used a combination of domain dictionary pre-annotation and multi-round annotation to annotate more than 340,000 words of aquatic animal disease corpus,and then built the aquatic animal disease corpus.The corpus was then constructed.(2)Proposed a text level aquatic animal disease entity relationship joint extraction method based on deep learning.Aiming at the problem of error propagation in traditional pipeline entity relation extraction method,this paper proposes a joint entity relation extraction method.This method firstly takes BERT as the input,annotates the text with BERT+BiLSTM+ATT+CRF model,and finally extracts the text with triples based on user-defined extraction rules.(3)A knowledge graph for aquatic animal disease diagnosis is constructed.In this paper,we use the BERT + BiLSTM + Attention + CRF model with fused label embedding method to complete the joint extraction of entity relationships in the field of aquatic animal diseases,and select the Neo4 j graph database to complete the knowledge storage of the extracted entity triples,and finally realize the construction of the knowledge graph of aquatic animal disease diagnosis.