Automatic classification of patents oriented to TRIZ: a case study on large aperture optical elements

CORC > 文献情报中心 > 中国科学院文献情报中心 > 中国科学院成都文献情报中心 > 信息技术部

	Automatic classification of patents oriented to TRIZ: a case study on large aperture optical elements
	Hu ZY(胡正银); Zhang X(张娴); Zhang X(张娴); Hu ZY(胡正银)
	2014
会议名称	4th global tech mining conference
会议日期	2014.9.2
会议地点	leiden,netherland
关键词	Topic model automatic classification SAO clumping semantic knowledge representation patent analysis
英文摘要	most of the current available patent classification systems are too general to triz users who pay more attention to find analogous patents in other fields that have solved the similar technical problems by using the same solutions. in this paper, we propose an approach to automatically classify patents oriented to triz applications based on a personalized classification schema. firstly, we construct a personalized classification schema in micro-meso-macro levels. micro-level is composed of subject-action-object (sao) extracted from patent text, meso-level problems solved and solutions used (p&s) topics generated by latent dirichlet allocation (lda) based on sao and macro-level technology domain generated by lda based on p&s topics. then, we choose an appropriate feature and classifier to preliminarily classify patents according to the personalized classification schema. finally, the classifier is optimized by smoothing imbalanced data and reducing features dimensions of sao. we evaluated the approach with large aperture optical elements (laoe) patent documents set as a case study. the results of case study showed that this approach can classify patents with high accuracy and speed and facilitate triz users to better utilize patents in medium size data set.
会议录	4th global tech mining conference
学科主题	信息组织与服务 ; 信息内容分析 ; 知识组织 ; 自然语言处理 ; 情报研究理论与方法
语种	英语
内容类型	会议论文
源URL	[http://ir.las.ac.cn/handle/12502/7240]
专题	文献情报中心_中国科学院成都文献情报中心_信息技术部
推荐引用方式 GB/T 7714	Hu ZY,Zhang X,Zhang X,et al. Automatic classification of patents oriented to TRIZ: a case study on large aperture optical elements[C]. 见:4th global tech mining conference. leiden,netherland. 2014.9.2.