支持技术创新的专利检索与分析

doi:10.11959/j.issn.1000-436x.2016055

摘要/Abstract

摘要：

介绍了目前专利检索和分析的主要研究工作，包括专利的可检索性、技术现状检索和相关性检索方法等，以及专利地图分析、新颖度分析和PatentDom专利分析框架等分析方法。最后基于深度学习的思想，讨论了新一代的支持技术创新的专利检索方法、专利论文检索方法以及专利趋势分析方法。

关键词: 专利, 专利检索, 专利分析, 深度学习

Abstract:

The main research work of patent search and analysis were summarizes.The patent search includes patentabil-ity search,prior art search,and query expansion.And the patent analysis includes patent map,novelty analysis,and a new analysis framework named PatentDom.Finally,based on the idea of deep learning,three new methods of patent search and analysis are put forward.

Key words: patent, patent search, patent analysis, deep learning

刘斌,冯岭,王飞,彭智勇. 支持技术创新的专利检索与分析[J]. 通信学报, 2016, 37(3): 79-89.

Bin LIU,Ling FENG,Fei WANG,Zhi-yong PENG. Patent search and analysis supporting technology innovation[J]. Journal on Communications, 2016, 37(3): 79-89.

图/表 15

表1

表2

表3

表4

表5

表6

图1

表7

表8

图2

表9

表10

表11

图3

图4

参考文献 41

[1]	State Intellectual Property Office of PRC.2014 key IP5 statistical data[EB/OL]. .
[2]	State Intellectual Property Office of PRC.2013 key IP5 statistical data[EB/OL]. .
[3]	State Intellectual Property Office of PRC.2012 key IP5 statistical data[EB/OL]. .
[4]	CHEN C . Searching for intellectual turning points:progressive know-ledge domain visualization[J]. PNAS, 2004,1011(Suppll): 5303-5310.
[5]	ERDI P , MAKOVI M , SOMOGYVARI Z , et al. Prediction of Emerg-ing technologies based on analysis of the US patent citation network[J]. Scientometrics, 2013,95(l): 225-242.
[6]	FUJII A , ISHIKAWA T . NTCIR-3 patent retrieval experiments at ULIS[C]// NII Test Collection for IR Systems-3. c2002:1-6.
[7]	FUJII A , ISHIKAWA T , KANDO N . Test collections for ptent-to-ptent rtrieval and ptent map generation in NTCIR-4 workshop[C]// The 4th International Conference on Language Resources and Eva uation. c2004:1643-1646.
[8]	FUJII A , ISHIKAWA T , KANDO N . Overview of the patent retrieval task at the NTCIR-6 workshop[C]// NII Test Collection for IR Sys-tems-6. Tokyo,Japan, c2007:359-365.
[9]	WU S , SUN J , TANG J . Patent partner recommendation in enterprise social networks[C]// WSDM. Rome,Italy, c2013:43-52.
[10]	JIN X , SPANGLER S , CHEN Y , et al. Patent maintenance recom-mendation with patent information network model[C]// ICDM. Vancouver,Canada, c2011:280-289.
[11]	MANNING C , RAGAVAN P , SCHUTZE H . An introduction to in-formation retrieval[M]. London: Cambridge University Press, 2009.
[12]	MAGDY W , JONES G . PRES:a score metric for evaluating re-call-oriented information retrieval applications[C]// SIGIR. Geneva,Switzerland, c2010:611-618.
[13]	HIDEO I , HIROKO M , YASUSHI O . Term distillation in patent re-trieval[C]// The ACL-2003 Workshop on Patent Corpus. c2013:41-45.
[14]	VERBERNE S , HONDT E D . Prior art retrieval using the claims section as a bag of words[C]// The Cross-language Evaluation Forum Conference on Multilingual Information Access Evaluation:Text Re-trieval Experiments. c2009:497-501.
[15]	VARMA M , VARMA V . Applying key phrase extraction to aid inva-lidity search[C]// International Conference on Artificial Intelligence and Law. Pittsburgh,PA, c2011:249-255.
[16]	KONISHI K . Query terms extraction from patent document for inva-lidity search[C]// NTCIR-5 Workshop Meeting. Tokyo,Japan, c2005.
[17]	MAHDABI P , ANDERSSON L , Keikha M , et al. Automatic refine-ment of patent queries using concept importance predictors[C]// SIGIR. Portland,USA, c2012:505-514.
[18]	TAKAKI T , FUJII A , ISHIKAWA T . Associative document retrieval by query subtopic analysis and its application to inva idity patent search[C]// CIKM, Washington,USA, c2004:399-405.
[19]	ADAMS S . Comparing the IPC and the US classification systems for the patent searcher[J]. World Patent Information, 2001,23(l): 15-23.
[20]	MAHDABI P , GERANI S , HUANG J X , et al. Leveraging conceptual lexicon:query disambiguation using proximity informat n for patent retrieval[C]// SIGIR. Dublin,Ireland, c2013:113-122.
[21]	GANGULY D , LEVELING L , MAGDY W , et al. Patent query reduc-tion based on pseudo-relevant documents[C]// CIKM. Glasgow,Scot-land,UK, c2011:1953-1956.
[22]	MAGDY W , JONES G . A study on query expansion methods for pa-tent retrieval[C]// PAIR. c2011:19-24.
[23]	KRESTEL R , SMYTH P . Recommending patents based on latent topics[C]// Recommender Systems. c2013:395-398.
[24]	MAHDABI P , CRESTANI F . Query-driven mining of citation net-works for patent citation retrieval and recommendation[C]// CIKM. Shanghai,China, c2014:1659-1668.
[25]	HIRONORI D , YOHEI S , et al. A patent retrieval method using a hierarchy of clusters at TUT[C]// NTCIR-5 Workshop Meeting. Tokyo,Japan, c2006.
[26]	BHATIA S , HE B , HE Q , et al. A scalable approach for performing proximal search for verbose patent search queries[C]// CIKM. Maui,HI,USA, c2012:2603-2606.
[27]	BASHIR S , AUBER A . Analyzing document retrievability in patent retrieval settings[C]// DEXA. c2009:753-760.
[28]	BASHIR S , AUBER A . Improving retrievability of patents in prior-art search[C]// ECIR. Dublin,Ireland, c2010:457-450.
[29]	SOOYOUNG O , ZHEN L , LEE W C , et al. CV-PCR:a context-guided value-driven framework for patent citation recommendation[C]// CIKM. San Francisco,CA,USA, c2013:2291-2296.
[30]	HUANG W , KATARIA S , CARAGEA C , et al. Recommending cita-tions:translating papers into references[C]// CIKM. Maui,HI,USA, c2012:1910-1914.
[31]	XUE X , CROFT W . Automatic query generation for patent search[C]// CIKM. Hong Kong,China, c2009:2037-2040.
[32]	JUN S H , PARK S , SIK J D . Technology forecasting using matrix map and patent clustering[J]// Industrial Management ＆ Data Systems. 2012,112(5): 786-806.
[33]	CHEN X , PENG Z , ZENG C . A co-training based method for chinese patent semantic annotation[C]// CIKM. Maui,HI,USA, c2012:2379-2382.
[34]	LIU D , PENG Z , LIU B . Technology effect phrase extraction in Chi-nese patent abstracts[C]// APWeb. Changsha,China, c2014:141-152.
[35]	DRAZIC M , KUKOLJ D , VITAS M , et al. Technology matchin of the patent documents using clustering algorithms[C]// The 14th IEEE International Symposium on Computational Intelligence and Infor-matics. c2013:405-408.
[36]	HASAN M A , SPANGLER S , GRIFFIN T , et al. COA:finding vel patents through text analysis[C]// SIGKDD. Paris,France, c2009:1175-1184.
[37]	ZHANG L H , LI L , LI T , et al. PatentLine:analyzing technology evolution on multi-view patent graphs[C]// SIGIR. Boston,Massachu-setts,USA, c2014:1095-1098.
[38]	ZHANG L H , LI L , LI T , et al. PatentDom:analyzing patent relation-ships on multi-view patent graphs[C]// CIKM. Shanghai,China, c2014:1369-1378.
[39]	TADURI S , YU H , LAU G , et al. Developing a comprehensive patent related information retrieval tool[J]. Journal of Theoretical and Ap-plied Electronic Commerce Research, 2001,6(2): 1-16.
[40]	BNEGIO Y , DUCHARME R , VINCENT P , et al. A neural probabilis-tic language model[J]. Journal of Machine Learning Research, c2009:1137-1155.
[41]	WANG M X , LU Z D , LI H , et al. GenCNN:a convolutional architec-ture for word sequence prediction[C]// ACL. c2015.

国家	第100万件	第200万件	第300万件	第400万件	第500万件
美国	100年	50年	25年	12年	5年半
中国	15年	7年半	2年半	1年	1年

会议	专利类型	时间范围	数量	主题数
NTCIR-3	日文专利	1998-1999	697 262	31
NTCIR-3	英文摘要	1995-1999	1 700 000	31
NTCIR-4	日文专利，英文摘要	1993-1997	1 700 000	103
NTCIR-5	日文专利，英文摘要	1993-2002	3 496 252	1 223
NTCIR-6	英文专利(USPTO)	1993-2002	1 315 470	3 221

QV	TV	p	q	MAP	P@10
QV₂	TV₃	t	i	0.279 4	0.390 3
QV₀	TV₃	t	i	0.270 1	0.348 4
QV₂	TV₃	a	i	0.268 8	0.364 5
QV₅	TV₃	a	w	0.263 7	0.361 3

算法	MAP	PRES	召回率
IPC扩展(第一声明)	0.129 3	0.514 0	0.606 7
IPC扩展(所有声明)	0.144 5	0.491 1	0.562 4
Ganguly^[^21]	0.127 8	0.460 4	—
Magdy^[^22]	0.139 9	0.486 0	—

算法	MAP
算法	平均准确率	提升（+）/下降（-）
BM25	0.062	-51.2%
LM	0.127	—
LDA	0.134	+5.5%
DMR	0.143	+12.6%
LM-DMR	0.164	+29.1%
LM-LDA	0.177	+39.4%