CPC/IPC classification

Every patent application is assigned codes locating it in a hierarchical technology taxonomy - the international IPC and the Cooperative Classification built on top of it. The codes are how patent corpora become technology statistics.

Evidence from this slice: Google Patents Public Data publications include 'CPC/IPC classifications' alongside titles, abstracts, claims, assignees and citations; USPTO PatentsView bulk TSVs cover 'bibliographic data, CPC/IPC/WIPO classifications, citations and full text from 1976 onward'; and USPTO Patent Public Search tags include 'cpc' for fielded and advanced searching.

Without classification codes, patent counts measure paperwork volume, not technology activity. With them, semiconductor or battery innovation becomes countable.

• Technology-field filtering is the core use case. Isolating electrical-machinery filings from consumer-product clutter requires the codes to be present and populated.
• Multiple schemes mean mapping work. Sources expose CPC, IPC and WIPO codes together; decide which hierarchy your analysis standardizes on before merging sources.
• Codes enable cross-office comparison. The same invention class filed in different jurisdictions shares a taxonomy, which is what makes office-to-office benchmarks possible.

1. Confirm codes ship with the bulk product. PatentsView includes classifications in its TSVs from 1976 onward; a source making you query one patent at a time is not equivalent.

1. Check which schemes are covered. CPC, IPC and WIPO labels serve different workflows - Google Patents and PatentsView both carry multiple.

1. Expect multiple codes per publication. One patent typically holds several classifications; counting publications per code needs the many-to-many link table, not a single column.

1. Support fielded search. Patent Public Search's 'cpc' tag enables advanced searching - verify the same filtering exists in any commercial alternative.

1. Combine with family identifiers when counting inventions across offices rather than documents.

Patent-analytics concepts used alongside classification codes: - inventor disambiguation — Resolving which patent records belong to the same real-world inventor despite inconsistent name spellings, turning raw grant text into analysable person-level entities. - ip statistics by technology field — Patent filing and grant counts disaggregated by technology field and issuing office, published by WIPO's statistics center. - patent family id — The field linking patent publications that describe the same invention filed in different offices.