面向水产养殖领域的对虾知识图谱云平台设计与实现

Design and Implementation of Shrimp Knowledge Graph Cloud Platform for Aquaculture Field

摘要: 针对互联网中存在的松散型、碎片化和难整合的水产养殖领域知识现状，将知识图谱应用于对虾养殖领域，面向多元异构数据源进行知识抽取，采用BI-LSTM-CRF模型进行命名实体识别、TextCNN模型进行关系识别、Neo4j数据库存储获取的知识数据，建立基于SpringBoot框架的对虾知识图谱云服务平台，将分散的对虾知识有效整合为一个规范化、标准化和系统化的知识库，并采用SparkMlib的朴素贝叶斯分类算法完成问题模板的匹配，实现基于知识图谱的对虾智能检索、智能推荐、智能问答和疾病辅助诊断等功能，为养殖户、企业和科研人员提供便捷、有效和系统化的对虾领域知识。

Abstract: The independence, fragmentation, and looseness of aquaculture knowledge contents on the internet lead aquaculture hard to search and obtain integrated and accurate knowledge. For this purpose, a knowledge graph cloud platform of shrimp in the aquaculture field based on knowledge graph and SpringBoot framework is established. In which, the scattered knowledge of shrimp is effectively integrated into a standardized, and systematic knowledge. The platform includes four parts: the Bidirectional Long and Short Term Memory Network Conditional Random Fields (BI-LSTM-CRF) model is used for named entity recognition for the sake of extracting the knowledge from multiple data sources. The Text Convolutional Neural Network (TextCNN) model is used for relationship recognition; Neo4j database is used to store the acquired knowledge. The intelligent question answering and the intelligent search and information recommendation of aquaculture knowledge based on knowledge graph are realized by naive Bayesian classification algorithm to complete the matching of question templates. Our study provides users with a good environment for knowledge learning and using.