The International Patent Classification (IPC) serves to classify the technical content of a patent document. The complete IPC can be found on the website of the World Intellectual Property Organization (WIPO).

The classification symbol is made up of a letter denoting the IPC section (eg B), followed by a number (two digits) denoting the IPC class (eg B60). Optionally, the classification can be followed by a sequence of a letter denoting the IPC subclass (eg B60N), a number (variable, 1-3 digits) denoting the IPC main group (eg B60N2), a forward slash "/", and a number (variable, 1-3 digits) denoting the IPC subgroup (eg B60N2/28). This makes the IPC symbol an alphanumerical code with no less than 11 characters.

PATSTAT table tls209_appln_ipc contains more than 80 million rows, so any search based on IPC expressed in alphanumerical terms can be very penalizing in computational terms. For this purpose we have reformulated tls209_appln_ipc just keeping the appln_id and ipc_symbol_class information, but the latter expressed in a numerical form. Also a concordance table in order to recover the original alphanumerical form has been built up.

At CEMI we built in CEMIPAT a modified table from PATSTAT original database and a complementary one:

• IPC numerical coding concordance table (CSV file, 340Kb)

• MySQL script for uploading ipc_coding.csv and creating tls209_appln_ipc_bis (SQL, 1Kb)