How to do an effective patent search
4 likes1,746 views
Here are the key steps to do an effective patent search in 3 sentences or less: Define the scope of your search by determining the relevant technology area or competitor. Identify relevant patent databases considering their coverage. Use both keyword and classification searches along with operators to effectively search multiple databases, as keyword searches alone may only find 10-20% of relevant patents.
How to do an effective patent search
- 1. How to do an effective Patent Search Björn Jürgens E-Mail: bjorn.jurgens@gmail.com Twitter: @benlogo © 2011 1
- 2. First of all… • Define Scope of Search for the Technology Watch Activity! – State of the Art of a specific sector? => keyword and classification search – Competitor Watch? =>name search • Then identify relevant patent databases (national or multinational, looking at the data coverage of each database)! 2
- 3. Keyword Search vs. Classification Search • Main downside of keyword searching are: – Poor translations: for example probably 40% of US patents are filed from abroad and not always well translated. – Misleading titles: describing something vulgar which sounds like something complex – Too many synonyms: especially in the realm of chemistry, a compound may have several names, depending on the country or field of use It is estimated that Keyword – No Spelling Standards (American English vs. British searches alone English) will find – Errors and Omissions: obvious mistakes in titles and names probably only (no content control in most databases!) 10-20% of – Date Limitations: Most databases only go back about 30 relevant years for searchable abstracts or text. Before that, there is patents! usually only patent classification. 3
- 4. Keyword Search vs. Classification Search • But a Classification only search also has its downsides: – there is no classification which adequately identifies the search object – there is only a classification which is too general (and therefore gives too many irrelevant results…) – the classification is behind the technology development • THEREFORE: – Intelligent use of keywords combined with classifications is the principal way to go – A thorough patent search will consist of dozens of search strategies involving class/keyword and class/class searches. 4
- 5. Patent Classifications • Assigned by Patent Examiners in the Patent Offices, nor by the inventors! • Each patent is classified at least with 1 clas.symbol => Still sometimes the classification is subjective and incomplete (depends on the patent examiner)! => Other problems: o Slow revision cycles: Sometimes classification systems do not keep up in time with technology development (and appearance of new technology that needs classification) 5
- 6. Patent Classifications • Main Patent Classifications: – International Patent Classification (IPC) - 70 000 groups – European Patent Classification (ECLA) - 150 000 groups – US Patent Classification - 150 000 groups – Japan Patent Classification (FI) - 190 000 groups 6
- 7. IPC and ECLA Classifications They are the most widely used patent classification systems! Alphanumerical Classification (combines numbers and characters): • Sections are the highest level of IPC and ECLA. • Each section is subdivided into classes. • Classes comprise one or more subclasses. • Each subclass is broken down into main groups, which can contain multiple subgroups. Differences • ECLA has many more subdivisions than the IPC. • These internal subgroups start with the symbol of an IPC group, followed by a combination of letters and numbers. 7
- 8. Example: IPC Classification for “Socks for sweaty feet”: A43B 17/10 8
- 9. Example: ECLA Classification for “Socks for sweaty feet”: A43B 17/10 3 more subgroups! 9
- 10. How to determine the right Classification? Hints and Tips • Using the “fishing” technique: – Search by keyword, company, or inventor in the Patent Database for related inventions, and look at the classifications for those patents you think are similar to your idea 10
- 11. 11
- 12. 12
- 13. How to determine the right Classification? Hints and Tips • Using classification catchword index searching interfaces: – WIPO IPC catchword index – ESPACENET ECLA catchword index – Different approach of these classification search engines: • WIPO: static matching • ESPACENET: statistical approach (more precise results!) 13
- 14. WIPO IPC catchword index ( http://www.wipo.int/ipcpub/ ): 14
- 15. WIPO IPC catchword index ( http://www.wipo.int/ipcpub/ ): 15
- 16. ECLA catchword index (http://worldwide.espacenet.com/eclasrch/ ): 16
- 17. How to determine the right Classification? Using free available Classification Software Tools: • TACSY: Search for relevant classification symbols in natural language – Natural language search of IPC: you need not know precise terminology – Takes shorter keyword combinations up to 15-20 words – Results given down to subgroups – http://www.wipo.int/tacsy/ • IPCCAT: Assistance for classification of patent applications – Takes texts up to 200 words (e.g. abstracts); document upload – Based on artificial neural network – http://www.wipo.int/ipccat 17
- 20. Using Operators • Most Patent Databases (and databases in general) work with operators, – either by connecting data fields (usually named advanced search) – or by connecting the keywords in a free text search interface (usually named “expert search”) • “Have to know” Operators: – Boolean Operators – Wildcards (Truncation, etc.) – Other Operators (Proximity, etc.) 20
- 21. Boolean Operators Most Some Asian Some Meaning Databases Databases Iberoamerican Databases OR + O GROUPING OPERATOR: BROADEN search and retrieve records containing any of the words it separates. For example: Looking up “COLGATE OR PHILIPS” as an asignee, retrieves documents which were filed either from Colgate or Phillips or both together. AND * Y COMBINING OPERATOR: NARROW search and retrieve documents containing ALL of the keywords it separates. For example: Looking up “COLGATE AND HENKEL” only retrieves documents which contain BOTH names. NOT - NO EXCLUDING OPERATOR: NARROW search and retrieve records that do not contain the term following it. For example: “COLGATE NOT PALMOLIVE” retrieves documents where COLGATE appears without PALMOLIVE. 21
- 22. Using Wildcards and other Operators Most Some Asian Some USPTO Meaning Databases Databases Iberoamerican Databases OPEN TRUNCATION: BROADEN search and stands for * + $ a string of caracters of any lenght For example: MICRO* will retrieve documents with MICROSOFT, MICROFILM, MICROPROCESSOR, etc. LIMITED TRUNCATION: BROADEN search and stands ? ? for zero or one carácter For example: CAR? will retrieve documents with CARS, but not CARSHARING, etc. COMPOUND TERMS: NARROW search and only “” “” “” “” retireves documents which the compound term. For example: “Philips Sensiflex” just retrievs documentos with the exact word order. PARENTESIS,useful for combining keywords with different () () () () boolean operators. For example: TOOTHBRUSH AND (ROTAT* OR 22 INTERACTIVE)
- 23. Using Operators and Wildcards – Hints and Tips In most free of cost databases there are a number of restrictions on the use of wildcards : • Wildcards cannot be followed by an alphanumeric character (colo?r is not allowed) • Wildcards can only be used in the "Title ", "Title or abstract ", "Inventor " or "Applicant " fields • There must be at least two alphanumeric characters preceding a ?. A maximum of three truncation symbols is then allowed (for example ca??? will retrieve call, cart, card, care, cable, etc.) • There must be at least three alphanumeric characters preceding a * symbol (co* is not allowed). 23
- 24. Using Operators and Wildcards – Hints and Tips • Input Fields in most databases are NOT CASE SENSITIVE! • Usually NO Wildcards needed when searching in (automatic wildcard) – classification fields – application number – priority number – publication number – publication date • You can retrieve all the documents having a particular country code simply by entering the country code (eg ES) in the number field. • You can retrieve all the documents published a particular year, simply by entering the first 4 digits of the date which corresponds to the year (for example 2010) 24
- 25. 25
- 26. Name Search (Applicant and/or Inventor Fields) Hints and Tips • Names of Persons (Inventors) are often not spelled correctly, especially if they are non English names! • Do not use non English characters (like accents, umlaut, etc.) • Try to use truncation – Valentin Nuñez = nunez v* • Try to include all possible variations of the names and connect them with the OR operator – Pedrosa-Rivas V* = pedrosa rivas V* OR pedrosarivas V* – O'Connor G = O'Connor, G OR OConnor, G. 26
- 27. Name Search (Applicant and/or Inventor Fields) Hints and Tips • Names of Applicants (Company Names) are often abbreviated or some parts of the name translated to English (or not) • For example patents of the same Spanish university can include: – UNIVERSITY OFGRANADA – UNIVERSIDAD DE GRANADA – UNIV GRANADA • Try to include all possible variations of the names and connect them with the OR operator! 27
- 28. Country Search • If you want identify patents published in a certain country, restrict your search with country codes! • List of Country Codes (Espacenet=> Help=> Glossary=>Country Codes): http://t1.espacenet.com/help?locale=en_T1&method=handleHelp Topic&topic=countrycodes 28
- 29. Patent Searching - Conclusion • There is no single best way to search patents! • Systematic keyword searching! – involves looking up narrow, broad, and related terms, grouped together in proximity and related to other groups of terms expressing functionality or application. – the terms must then be searched in rotation including every conceivable permutation and combination. • On finding relevant patents, check all classifications (USPTO, ECLA, IPC) • Check all cited and citing patents for relevant patents found. • On finding relevant patents, check patents by the same inventor or company. • Do a search in full text if possible! 29
- 30. Thank you very much! Contact Details: Björn Jürgens E-Mail: bjorn.jurgens@gmail.com Twitter: @benlogo 30