A framework for Semantic Web of Patent Information
Advising professor: Dr. Hei-Chia Wang Speaker: Yung Chang Chi Time: 2016/06/29
Outline
• Introduction • Literature review • Research Methods • PATExpert • Semantic Web • Future WorkOutline
• Introduction • Literature review • Research Methods • PATExpert • Semantic Web • Future Work 3Introduction
Yung Chang Chi
National Cheng Kung University. Tainan, Taiwan
Department of Industrial and Information Management Doctoral Candidate .
Master of Engineering: University of South Australia, Australia Adelaide (2000-2001) Master of Technology Law: National Kaohsiung First University of Science and Technology, Kaohsiung, Taiwan (2006~2009)
Introduction
• Patents are important knowledge sources for industrial
research and product development because of their
innovation and practicability.
• In recent years, patent analysis increased in
importance in high-technology management as
process of innovation became more complex.(Y.
Liang, R. Tan, and J. Ma, 2008).
Introduction
• The judgements of patent infringement, unlike the
patent documents, can be mined using text mining
techniques, since the judgements are legal
documents.
• The judgements can be transformed into patterns
by content analysis, and readers can easily access
them the same way as reading newspapers to
understand the key points and issues in dispute.
Judgements of patent infringement
• MICROSOFT CORP. v. AT&T CORP.
• certiorari to the united states court of appeals for the federal circuit
• No. 05-1056. Argued February 21, 2007--Decided April 30, 2007
• It is the general rule under United States patent law that no infringement occurs when a patented product is made and sold in another country. There is an exception. Section 271(f) of the Patent Act, adopted in 1984, provides that infringement does occur when one "suppl[ies] ... from the United States," for "combination" abroad, a patented invention's "components." 35 U. S. C. §271(f)(1). This case concerns the applicability of §271(f) to computer software first sent from the United States to a foreign manufacturer on a master disk, or by electronic transmission, then copied by the foreign recipient for installation on computers made and sold abroad.
• AT&T holds a patent on a computer used to digitally encode and compress recorded speech. Microsoft's Windows operating system has the potential to infringe that patent because Windows incorporates software code that, when installed, enables a computer to process speech in the manner claimed by the patent. Microsoft sells Windows to foreign manufacturers who install the software onto the computers they sell. Microsoft sends each manufacturer a master version of Windows, either on a disk or via encrypted electronic transmission, which the manufacturer uses to generate copies. Those copies, not the master version sent by Microsoft, are installed on the foreign manufacturer's computers. The foreign-made computers are then sold to users abroad.
Outline
• Introduction • Literature review • Research Methods • PATExpert • Semantic Web • Future Work 9• J. Michel, and B. Bettels, “Patent citation analysis: a closer look at the basic input data from patent search reports”,
Scientometrics. 2001, pp.185-201. Vol.51. no. 1.
• I. S. Kang, S.H. Na, J. Kim, and J.H. Lee, Cluster-based patent retrieval. “Information Processing & Management”. 2007, pp.1173-1182.43(5).
• [7] J.H. Kim, and K.S. Choi, Patent document
categorization based on semantic structural information. “Information processing & Management”. 2007, pp.1200-1215.43(5).
Outline
• Introduction • Literature review • Research Methods • PATExpert • Semantic Web • Future Work 11Patent documents analysis
• Base on the collected patent documents and the subject-action-object (SAO) structures extracted by using Natural Language Processing (NLP).
• NLP tools will be used for build a set of SAO structures from the collected patents.
• Multidimensional scaling (MDS) is a statistical technique used to visualize similarities in data.
• Patent documents in different fields have different key issues that trigger different Multidimensional scaling.
• The paper will design a new algorithm to identify which particular patent field shall correspond to what extent of scaling.
Patent documents analysis
Patent infringement verdict content
analysis
• The text in the patent infringement judgements is important because that is where the meanings are.
• Content analyses commonly contain six steps : Design Unitizing Sampling Coding Drawing inferences Validation 15
Comparative Analysis
19 Isolated Patent Comparative Analysis Patent Documents Patent Infringement ContentWordNet
• WordNet was created in the Cognitive
Science Laboratory of Princeton University
under the direction of psychology professor
George Armitage Miller starting in 1985.
Word Net
Semantic Web similarity
comparison table
Patent Engineer marked similarity No include Semantic net income of similarity Include semantic net income of similarity 0.9 0.6 0.7 0.8 0.7 0.8 0.7 0.5 0.6 0.6 0.4 0.5 0.5 0.4 0.5 0.4 0.6 0.7 0.3 0.5 0.6 0.2 0.4 0.5 0.1 0.1 0.2 0.0 0.2 0.3Outline
• Introduction • Literature review • Research Methods • PATExpert • Semantic Web • Future work 23PATExpert
• The European project PATExpert,
(Advanced Patent Document Processing
Techniques), coordinated by Barcelona
Media (BM), successfully accomplished the
objectives settled, after the pre-established
30 months (from February 2006 to July
PATExpert
Outline
• Introduction • Literature review • Research Methods • Search Engine • Semantic Web • Future workSemantic Web
27
J. Hebeler, M. Fisher, R. Blace and A. Perez-Lopez, “Semantic Web Programming “Wiley Publishing, Inc.2009.
Constructing Semantic Web
• Reasoners: Reasoners add inference to the
Semantic Web. Inference creates logical
additions that offer classification and
realization.
• The semantic web components of Rules
engines support inference typically beyond
what can be deduced from description logic.
Outline
• Introduction • Literature review • Research Methods • Search Engine • Comparative Analysis • Future work 31The next steps will be to employ the image recognition functions to identify drawings and pictures.
33
