摘要
基于当前装备故障诊断的现状,依据在装备维修手册、装备履历书以及装备管理信息系统中存在大量的装备故障和维修经验等数据,结合装备故障文本的特点,提出了一种融合词性、语义及词序因子的故障文本相似度计算方法。该方法将装备故障文本中词汇的词性、语义及位置关系相联系,在余弦公式的基础上,通过文本中的词汇之间的相似度与词性权重的关联关系,改进相似度计算方法,并引入词序相似度进一步优化文本相似度。实验表明,所提出的方法较其他方法有更好的精确率和召回率,有效提高了装备故障文本的匹配效果。
Based on the current situation of equipment fault diagnosis,this paper took the advantage of the data of equipment fault and maintenance experience in the equipment maintenance manual,equipment resume and equipment management information system,and finally presented a method for calculating the fault text similarity by fusing parts of speech,semantics and word order factors.Based on the cosine formula,the similarity between the words and the weight of part of speech was used to improve the similarity calculation method,and the word order similarity was introduced to optimize the text similarity.Experimental results show that this method has better accuracy and recall than other methods,and improves the matching effect of equipment fault text effectively.
作者
祖月芳
凌海风
吕永顺
ZU Yuefang;LING Haifeng;LYU Yongshun(College of Field Engineering, Army Engineering University, Nanjing 210004, China)
出处
《兵器装备工程学报》
CSCD
北大核心
2021年第11期204-208,共5页
Journal of Ordnance Equipment Engineering
关键词
装备故障文本
词向量
词性
语义
词序相似度
文本相似度
匹配算法
equipment fault text
word vector
part of speech
semantics
word order similarity
text similarity
matching algorithm
