Design And Implementation Of Knowledge Graph System For Prawn Culture Domain

With the popularization of artificial intelligence and Internet technology in the field of prawn,the data in the field of prawn has gradually increased and the structure has become more complex."Smart prawn culture" will become the development trend of the prawn breeding industry in the future,which can provide the public with massive amounts of information in the prawn field.Due to the low degree of resource sharing in the prawn field,it is difficult for farmers to search the required information efficiently and conveniently.While,the knowledge graph technology can well integrate and visualize the knowledge of the prawn field.Therefore,this article takes the construction of a knowledge graph in the field of prawn culture as the research goal,and conducts in-depth research on the realization technologies of the prawn knowledge graph service system.The specific work of this paper contains the following four points:Firstly,constructing the corpus of prawn domain.Based on the Web Magic crawler framework,this paper uses Baidu Encyclopedia and the prawn domain information website as the source of the prawn corpus data,crawls prawn domain related data for named entity recognition experiment,analyzes the characteristics of the text in the prawn culture domain,and then designs the corresponding rules of prawn entity annotation by using BIOES labeling method.Using the Han LP tool to conduct the data of prawn field,such as Chinese word segmentation and part of speech tagging.After the word vector training,a corpus of the prawn field will be generated and used in subsequent research work.Secondly,research on a named entity recognition model in the prawn field.This paper uses the Bi LSTM-CRF named entity recognition model to identify 44 types of prawn farming field entities from the prawn texts,and then uses HMM,CRF and Bi LSTM three kinds of models to conduct comparative experiments with this model.The results show that the F1 value of this model was reached 90.50% in the manually labeled shrimp test corpus.Its entity recognition effect is markedly better than the other three kinds of traditional model,which can effectively identify named entities in the prawn field.Thirdly,constructing a knowledge graph in the field of shrimp culture.At present,there is even less open source knowledge graph involving the field of prawn.Therefore,this article uses the knowledge graph construction technology.In order to extract the entities and relationships that need to construct the knowledge graph,this article uses web crawler technology to extract the isomerization prawn domain knowledge from the Internet and prawn expert monographs.Then use the knowledge extraction technology to transform it into structured data.At last,the Neo4 j graph database is used as the knowledge storage tool of the prawn knowledge graph,and initially realize the visualization of the prawn domain knowledge graph.Finally,the knowledge graph service system of prawn culture was designed and implemented.Comprehensive application of the above research results,this article constituted a knowledge graph service system in the field of prawn culture based on the Spring-Boot framework,which can support the following functions: basic information query of shrimp,visualization of shrimp knowledge map,shrimp entity query,shrimp text named entity recognition,Intelligent search for shrimp information.The purpose of constructing a knowledge map service system in the field of shrimp farming is to provide users with a more professional,systematic,intuitive and intelligent sharing service of knowledge map information for shrimp farming,which has certain research value for promoting the development of intelligent information technology in the shrimp farming industry.