基于知识图谱的羊群疾病问答系统的构建与实现

doi:10.13300/j.cnki.hnlkxb.2023.03.008

首页 > 过刊浏览>2023年第42卷第3期 >63-70. DOI:10.13300/j.cnki.hnlkxb.2023.03.008

基于知识图谱的羊群疾病问答系统的构建与实现
DOI:
                        10.13300/j.cnki.hnlkxb.2023.03.008
                    
CSTR:
                        [cstr]
                    
作者:
                        杨喆1,2杨喆
华中农业大学信息学院/农业农村部智慧养殖技术重点实验室/农业智能技术教育部工程研究中心/ 湖北省农业大数据工程技术研究中心，武汉430070;华中农业大学动物科学技术学院/农业动物遗传育种与繁殖教育部重点实验室，武汉430070
在期刊界中查找
在百度中查找
在本站中查找
许甜1许甜
华中农业大学信息学院/农业农村部智慧养殖技术重点实验室/农业智能技术教育部工程研究中心/ 湖北省农业大数据工程技术研究中心，武汉430070
在期刊界中查找
在百度中查找
在本站中查找
靳哲1靳哲
华中农业大学信息学院/农业农村部智慧养殖技术重点实验室/农业智能技术教育部工程研究中心/ 湖北省农业大数据工程技术研究中心，武汉430070
在期刊界中查找
在百度中查找
在本站中查找
孔玫1孔玫
华中农业大学信息学院/农业农村部智慧养殖技术重点实验室/农业智能技术教育部工程研究中心/ 湖北省农业大数据工程技术研究中心，武汉430070
在期刊界中查找
在百度中查找
在本站中查找
李国亮1李国亮
华中农业大学信息学院/农业农村部智慧养殖技术重点实验室/农业智能技术教育部工程研究中心/ 湖北省农业大数据工程技术研究中心，武汉430070
在期刊界中查找
在百度中查找
在本站中查找
杜小勇1,2杜小勇
华中农业大学信息学院/农业农村部智慧养殖技术重点实验室/农业智能技术教育部工程研究中心/ 湖北省农业大数据工程技术研究中心，武汉430070;华中农业大学动物科学技术学院/农业动物遗传育种与繁殖教育部重点实验室，武汉430070
在期刊界中查找
在百度中查找
在本站中查找

                    
作者单位:1.华中农业大学信息学院/农业农村部智慧养殖技术重点实验室/农业智能技术教育部工程研究中心/ 湖北省农业大数据工程技术研究中心，武汉430070;2.华中农业大学动物科学技术学院/农业动物遗传育种与繁殖教育部重点实验室，武汉430070
作者简介:杨喆，E-mail：1148713279@qq.com
通讯作者:杜小勇，E-mail：duxiaoyong@mail.hzau.edu.cn
中图分类号:TP391
基金项目:国家自然科学基金项目（ 31872978）

Construction and application of knowledge graph of sheep & goat disease

Author:

YANG Zhe ^{^1,2}
YANG Zhe
College of Informatics， Huazhong Agricultural University/ Key Laboratory of Smart Farming for Agricultural Animals， Ministry of Agriculture and Rural Affairs/ Engineering Research Center of Agricultural Intelligent Technology， the Ministry of Education/Hubei Engineering Technology Research Center of Agricultural Big Data， Wuhan 430070， China;College of Animal Sciences & Technology， Huazhong Agricultural University/Key Laboratory of Agricultural Animal Genetics，Breeding and Reproduction of Ministry of Education，Wuhan 430070，China
在期刊界中查找
在百度中查找
在本站中查找
XU Tian ^¹
XU Tian
College of Informatics， Huazhong Agricultural University/ Key Laboratory of Smart Farming for Agricultural Animals， Ministry of Agriculture and Rural Affairs/ Engineering Research Center of Agricultural Intelligent Technology， the Ministry of Education/Hubei Engineering Technology Research Center of Agricultural Big Data， Wuhan 430070， China
在期刊界中查找
在百度中查找
在本站中查找
JIN Zhe ^¹
JIN Zhe
College of Informatics， Huazhong Agricultural University/ Key Laboratory of Smart Farming for Agricultural Animals， Ministry of Agriculture and Rural Affairs/ Engineering Research Center of Agricultural Intelligent Technology， the Ministry of Education/Hubei Engineering Technology Research Center of Agricultural Big Data， Wuhan 430070， China
在期刊界中查找
在百度中查找
在本站中查找
KONG Mei ^¹
KONG Mei
College of Informatics， Huazhong Agricultural University/ Key Laboratory of Smart Farming for Agricultural Animals， Ministry of Agriculture and Rural Affairs/ Engineering Research Center of Agricultural Intelligent Technology， the Ministry of Education/Hubei Engineering Technology Research Center of Agricultural Big Data， Wuhan 430070， China
在期刊界中查找
在百度中查找
在本站中查找
LI Guoliang ^¹
LI Guoliang
College of Informatics， Huazhong Agricultural University/ Key Laboratory of Smart Farming for Agricultural Animals， Ministry of Agriculture and Rural Affairs/ Engineering Research Center of Agricultural Intelligent Technology， the Ministry of Education/Hubei Engineering Technology Research Center of Agricultural Big Data， Wuhan 430070， China
在期刊界中查找
在百度中查找
在本站中查找
DU Xiaoyong ^{^1,2}
DU Xiaoyong
College of Informatics， Huazhong Agricultural University/ Key Laboratory of Smart Farming for Agricultural Animals， Ministry of Agriculture and Rural Affairs/ Engineering Research Center of Agricultural Intelligent Technology， the Ministry of Education/Hubei Engineering Technology Research Center of Agricultural Big Data， Wuhan 430070， China;College of Animal Sciences & Technology， Huazhong Agricultural University/Key Laboratory of Agricultural Animal Genetics，Breeding and Reproduction of Ministry of Education，Wuhan 430070，China
在期刊界中查找
在百度中查找
在本站中查找

Affiliation:

1.College of Informatics， Huazhong Agricultural University/ Key Laboratory of Smart Farming for Agricultural Animals， Ministry of Agriculture and Rural Affairs/ Engineering Research Center of Agricultural Intelligent Technology， the Ministry of Education/Hubei Engineering Technology Research Center of Agricultural Big Data， Wuhan 430070， China;2.College of Animal Sciences & Technology， Huazhong Agricultural University/Key Laboratory of Agricultural Animal Genetics，Breeding and Reproduction of Ministry of Education，Wuhan 430070，China

Fund Project:

摘要

图/表

访问统计

参考文献

相似文献

引证文献

资源附件

文章评论

摘要:

为解决羊群疾病检索过程中出现的大量冗余数据及检索后仍需人工挑选准确答案造成的资源浪费，本研究通过以下3个步骤构建基于知识图谱的羊群疾病问答系统：（1）通过爬虫获取数据，人工提取部分信息，再进行自动化信息抽取，在命名实体识别任务中使用双向长短期记忆循环神经网络Bi-LSTM模型，并添加注意力机制提高识别效率，然后使用BIO规则进行实体标注，完成信息抽取，将数据融合后存储在Neo4j图数据库中，构建羊群疾病知识图谱。（2）针对属性映射，构建Bert-softmax模型；根据用户提问，采用Bert模型计算问句和属性的语义相似度，并通过softmax算法进行归一化处理，返回合适答案给用户，实现羊群疾病问答系统算法设计。（3）构建羊群疾病诊断平台，使用Bootstrap、Echarts、Vue组件实现羊群疾病问答系统的可视化，利用Python语言包含的flask框架搭建后台，封装疾病信息，通过web前端呈现给用户，并于后端建立连接，实现数据之间的交互。试验结果显示，基于Bi-LSTM + Attention + CRF模型实体识别的F₁值为83.16%，构建的知识图谱包含实体4 576个，实体关系超13 000条；问答系统添加了预训练模型Bert，对问题识别的F₁值为85.24%。结果表明，该系统实现了对羊群疾病的防治措施等多类问题进行快速检索和精准回答，可以辅助养殖人员在面临羊群疾病时进行生产决策。

关键词:疾病诊断;知识图谱;问答系统;Neo4j;Bert;智慧养殖

Abstract:

In order to solve the problem of a large amount of redundant data in the retrieval process of sheep disease and the waste of resources caused by manual selection of accurate answers after retrieval， this study constructed a question-and-answer system based on the knowledge graph of sheep & goat disease through the following three steps：（1） The data was obtained through web crawlers and some is manually extracted， automated information extraction was carried out using the bidirectional long short-term memory recurrent neural network （Bi-LSTM） model with an attention mechanism for improved recognition efficiency in the named entity recognition task. The entity annotation was performed using the BIO rule to complete the information extraction. The data was then integrated and stored in the Neo4j graph database . （2） For the attribute mapping， we constructed the Bert-softmax model； according to the user’s question， the Bert model was used to calculate the semantic similarity between the question and the attribute to determine the user’s intention， then the softmax algorithm was used to for normalization， finally， the most suitable answer was found and fed back to the system. （3） We built a sheep & goat disease diagnosis platform using Bootstrap， Echarts， and Vue components to visualize the sheep disease question-and-answer system.We used flask framework included in the Python language to build a backend， encapsulate disease information， present it to users through the web frontend， and establish a connection on the backend to enable data interaction. The results in the study show that the F₁ value of entity recognition based on Bi-LSTM + Attention + CRF model is 83.16%， and the constructed knowledge graph contains 4 576 entities and more than 13 000 entity relationships. The pre-trained model Bert was added to the question answering system， and the F₁ value of problem recognition was 85.24%. The results indicated that the system can quickly retrieve and accurately answer various types of questions such as the prevention and control measures of sheep diseases， and assist the farmers to make production decisions when faced with sheep diseases.