摘要
[目的/意义]提出一种农业领域的本体构建方法(手工+机器学习法),构建番茄病虫害的领域本体,旨在为番茄病虫害信息检索系统、诊断系统等平台的开发提供支撑。[方法/过程]首先,利用叙词表、文献资料确定本体类及层次和属性,构建初始本体框架;其次,通过机器学习从文献资料和网页信息中抽取相关实例;最后,通过本体描述语言将本体类、属性和实例形式化。[结果/结论]提出的本体构建方法可以初步实现农业领域本体的半自动、高效构建,命名实体识别的准确率为82.51%,关系的抽取准确率达到了86.32%。
[Purpose/significance]The paper proposes a method of building agricultural ontology(manual+machine learning method)to construct the domain ontology of tomato diseases and insect pests,so as to provide support for the development of tomato diseases and insect pests information retrieval system,diagnosis system and other platforms.[Method/process]Firstly,it uses thesaurus and literature data to determine the ontology class,level and attribute,and build the initial ontology framework;secondly,it uses machine learning to extract relevant examples from literature and web information;finally,it uses ontology description language to formalize ontology class,attribute and instance.[Result/conclusion]The ontology construction method proposed can initially realize the semi-automatic and efficient construction of agricultural ontology.The accuracy rate of named entity recognition is 82.51%,and the accuracy rate of relation extraction is 86.32%.
作者
任妮
孙艺伟
鲍彤
郭婷
Ren Ni;Sun Yiwei;Bao Tong;Guo Ting(Institute of Agricultural Information,Jiangsu Academy of Agricultural Sciences,Nanjing Jiangsu 210014)
出处
《情报探索》
2021年第4期51-57,共7页
Information Research
基金
国家社会科学基金一般项目“面向乡村振兴的图书馆精准服务及实现机制研究”(项目编号:19BTQ032)
江苏省农业科技自主创新资金一类项目“高效设施种养大数据应用关键技术研发与示范”(项目编号:CX(18)2029)研究成果之一。
关键词
机器学习
番茄病虫害
领域本体
machine learning
tomato diseases and insect pests
domain ontology