Effective Patent : Making Sense of the Information Overload Daniel R. Cahoy Smeal College of Business Penn State University VALGEN Workshop January 20-21, 2011
Patent vs. Statistical Analysis Statistical Analysis: correlations with patent activity Impact of policy or factor Factors correlated with certain types of patents Sophisticated econometrics are commonly used Individual attributes or interaction not as important : characterizing an invention space Identify Patents Associated with Particular Firm, Industry, Nation, etc. Assess portfolio value in relation to other patents Identify top inventors and knowledge flow Assess a technology environment Foundational versus improvement patents Temporal evolution Network and semantic searching common Statistical Analysis informs
Fundamental Patent Concepts for Patents provide only negative right to exclude Owning a patent provides no freedom except as defense Ownership may not provide market power Products are covered by multiple patents Patents can have dramatically different scopes Identified only in claims Invalidity is important, but hard to assess factor True understanding of patent power requires invalidity assessment Applications may not issue as patents Application landscapes provide fuzzy future view
Forms: Claim Analysis as Holy Grail Detailed claim analysis is best way to understand patent rights Infringement and validity analysis often combined Very time consuming and expensive Arguable results (only binding when court decides) Impractical for large (or unknown) numbers of patents Useful as top-level assessment Semantic ranking of claims? Proxies for strong claims?
Forms Level 1: Patent Counting Number of patents in particular area Merely categorizing statistical data Technology comparisons Temporal comparisons Challenges Capturing the correct collection of patents Type I and type II errors Pre-analysis necessary Patent Office classifications imperfect Keyword or semantic searches Early terms may not be settled Connections between patents Shared citations How good do you need to be?
Ex: Patenting in 2 nd Generation Bioethanol Patents related to ethanol production from Biomass increasing dramatically Type I and type II errors Are keyword-based trends influenced by outside factors? Important patents unknown Source: Cahoy & Glenna (2009) [1] United States Patent & Trademark Office (USPTO) issued patents database (http://patft.uspto.gov/netahtml/pto/search-adv.htm) using a search string designed to identify all patents containing relevant terms that resided in plant and microorganism classes: (ccl/800/$ or ccl/435/$) and ethanol and (lignocellulos$ or cellulos$) and (fuel or fuels)
Ex: WIPO & EPO Energy Patent Assessment WIPO Alternative Energy Patent Activity Study IPC + keyword searching of patent apps. Additional characterization Ownership, citation freq., trends EPO/UNEP/ICTSD Study of Patents and Clean Energy PATSTAT Claimed priorities Both studies depend heavily on understanding of technology For relevant figures, see: http://www.wipo.int/export/sites/www/patentscope/en/te chnology_focus/pdf/landscape_alternative_energy.pdf http://documents.epo.org/projects/babylon/eponet.nsf/0/c c5da4b168363477c12577ad00547289/$file/patents_clean_ energy_study_en.pdf Source: WIPO Analysis Report (2010) Source: EPO Report (2010)
Forms Level 2: Ownership Who owns the patents in a particular field? Or what does a particular company own? Dominant firms May lead to other fields Somewhat accessible More than just formal assignments Software provides useful assist
Ex: Patent Consolidation Glenna & Cahoy (2009) Comparative trends in biofuel consolidation EPO and WIPO studies For relevant figures, see: Source: USPTO Issued Patents Database http://www.wipo.int/export/sites/www/patentscope/en/te chnology_focus/pdf/landscape_alternative_energy.pdf http://documents.epo.org/projects/babylon/eponet.nsf/0/c c5da4b168363477c12577ad00547289/$file/patents_clean_ energy_study_en.pdf Source: EPO Report (2010) Source: WIPO Analysis Report (2010)
Level 1 and 2 Resources National/Regional patent offices Full, accurate information, but lacking analytical tools Downloadable products (PATSTAT, USPTO) International Entities WIPO s PatentScope (applications only) OECD Patent Statistics Google Patent New project to make more USPTO data available Level 1 and 2 landscaping can be done for very lost cost (or free)
Types Level 3: Connections Relationship between patents in innovation space Citation or semantic connection Complementary technology Overlapping technology Connections can be hard to interpret Research on networks and information flow Weighted links Relevant citation networks All Backward Citation Links Patents Issued Nov.-Dec. 2007: Class 800/300.1 Source: USPTO Issued Patents 102/103 Backward Citation Links Patents Issued Nov.-Dec. 2007: Class 800/300.1 Source: USPTO Issued Patents
Ex: Biofuel Patent Interconnections Free IPVision search of citation links among cellulosic ethanol group Technology clusters Temporal But missing patent scope info Source: www.see-the-forest.com
Types Level 4: Important Patents Some patents are worthless Identifying high value patents depends on many factors Scope and Validity Current and future market Direct measures of valuation for individual patents Income valuation (licensing) Statistical measures Citation by other patents (particularly forward citation) Membership in families (local and international) Citation in standard Renewal Other correlated proxies Seminal research on valuable patents Allison et al
Level 3 and 4 Sources Data is free, but assessment requires more work Proprietary software can provide useful overview Thomson Innovation IPVision References: Patent Information User s Group http://wiki.piug.org/pages/viewpage.action?pageid=101 26064 World Patent Information (Elsevier) Patent Searching (Hunt, ed. Wiley 2007) Heavily manipulated data must still be subject to detailed assessment
for Clearance-type Landscapes 1. Spend substantial effort to construct pool of relevant patents Interview technology experts, revise, consult outside studies 2. Use level 3 and 4 analyses as a guide to find most important patents Several types of analysis to generate study group E.g., forward citation and families; semantic claim analysis 3. Conduct attorney-driven claim analysis on hot spots Drill down as necessary