Search for Tomorrow: The Patent Office’s Growing Reliance on Technology to Locate Prior Art Andrew Chin unc. edu IPSC 2007
Patent Office Automation 1984: Full-text search from 1976 (two terminals) 1991: Full-text search from : Desktop access 1999: EAST, WEST interfaces 2000: “Big transition” 2001: Full-text search from 1920 (OCR) 2005: USPTO completes move to Alexandria, leaving “shoes” behind
MPEP § General Search Guidelines –Text search is powerful, but rarely sufficient by itself –“Some combination of text search with other criteria, in particular classification, would be a normal expectation in most technologies.”
EAST/WEST Supported Queries Terms: –Keywords, phrases –Classes, subclasses Connectors: –Boolean operators –Proximity operators –Truncation (i.e., stemming) Field restrictions
Questions Does keyword searching tend to diversify prior art? –Diversity of classes/subclasses? Effect of subject matter variations? –Diversity of ages? Pre-1976 (1971) citations? Stratification in citability rate? Time decay in citability rate?
Data Available: –U.S. patents from 1976 Full text Citations –Image File Wrappers from 8/2004 Examiner’s Search Strategy & Results (ESSR) reports from 2006 (scanned) Not available: –Citations derived from search results –Search engine query logs
ESSR
Synthetic Approach Use Moby dictionary M (354,984 words) Impute (Citing, Cited) to keyword search if there is a word w є M such that: –w appears in the claims and detailed descriptions of both Citing and Cited –w appears in the claims and detailed descriptions of no more than 50 total patents
Growing Reliance on Keyword Search
Technological Classes Amenable to Keyword Search
Comments Stephen Kunin: –Keyword searching is more useful in “the chemical area where the terms are better defined” Nestor Ramirez, MPEP: –Examiners in mechanical arts tend to look at all the patents in the class; biotech and chemical examiners rely almost entirely on text search
Questions Does keyword searching tend to diversify prior art? –Diversity of classes/subclasses? Effect of subject matter variations? –Diversity of ages? Pre-1976 (1971) citations? Time decay in citability rate? –Overall diversity? Stratification in citability rate?
Classification Diversity: By Subject Matter
Keyword Search Preference for Post-1976 References
Citability
Alpha
Beta