摘要
为了更好地检索出用户满意的冲压件实例,将相关反馈过程与模块加入冲压件实例检索系统。给出了基于向量空间的冲压件实例相似度计算公式与反馈迭代算法,分两个阶段考察新旧实例间的相似度,其结果由用户判断其相关性,采用Ide-regu-lar算法进行反馈运算,该步骤可以不断循环以逼近较为满意的实例。采用基于离差信息的权重更新算法,综合用户检索偏好和实例自身信息计算其组合权重。结果表明,所提出的反馈计算方法有利于提高冲压件检索准确度。
In order to seek the satisfactory stamping parts cases,relevance feedback process and module are added into the case retrieval system.The calculation formula of stamping parts cases similarity degree and relevance feedback iterative method are presented.The results are calculated in a dual-step between the new part and old ones,are judged by the users to define its relativity,and are taken into the feedback operation by means of Ide-regular algorithm.This course can be repeatedly circled to approach to a rather satisfied case.Deviation information is adopted into the weight update algorithm,which can combine the users’ search preference and the self-information from cases.The results show that this feedback means has high effectiveness in the similarity retrieval system.
出处
《计算机工程与应用》
CSCD
北大核心
2011年第7期241-245,共5页
Computer Engineering and Applications
基金
国家高技术研究发展计划(863)No.2008AA04Z113
上海市科委基础研究重点资助项目(No.08JC1419400)~~
关键词
相关反馈
实例检索
相似度
权重
relevance feedback
case retrieval
similarity degree
weight