摘要
针对现有农业病虫害专家系统领域知识共享和重用困难、症状输入严重影响准确率等问题,研究了基于领域知识、推理知识和任务知识的本体构建技术。同时,提出了基于分词、改进的词义消歧算法、同位词消歧及BNF范式形式化等自然语言处理步骤将症状描述训练集转化为本体形式,从而使专家系统引入本体技术,并利用其优势极大地改进了系统的健壮性及准确率。
This paper studies the ontology construction technique based on domain,reasoning and task knowledge,considering the difficulties in knowledge acquisition,sharing and reuse,and the lower accuracy relating to symptom input of existing expert systems of aquaculture diseases.This paper presents a method to convert the training set of symptoms describing into the form of ontology which is composed of segmentation,improved semantic disambiguation,apposition disambiguation and BNF formalization definition,so as to be applied to guarantee the expert system efficiency and reliability.
出处
《农机化研究》
北大核心
2012年第1期41-45,共5页
Journal of Agricultural Mechanization Research
基金
国家科技重大专项课题(2009ZX03001-09-01)
天津市农业科技成果转化与推广项目(2009P106)
