Gap betweeen US patent counts & DocDB simple family counts

Here you can post your opinions, ask questions and share experiences on the PATSTAT product line. Please always indicate the PATSTAT edition (e.g. 2015 Autumn Edition) and the database (e.g. PATSTAT Online, MySQL, MS SQL Server, ...) you are using.
Post Reply

Posts: 2
Joined: Thu Oct 25, 2018 12:05 pm

Gap betweeen US patent counts & DocDB simple family counts

Post by Tobor » Thu Oct 25, 2018 3:44 pm

Dear all,

I'm currently concerned with counting the number of annually granted patents in the USA between 1973 and 2013. To do this, I restricted my analysis on patents, the application authority of which is 'US', which are of application kind 'A' (direct filing) and which are granted. In order to count distinct inventions only, I use the definition of the DocDB simple family. That is, in a first step, I count the number of patents that belong to the same DocDB family inside my retrieved data set. In a second step, I select the patent with the earliest application filing date from each DocDB family as a representative. I get two curves: one showing raw patent counts over the years (with all of the granted patents in that time frame), the second showing only one representative per DocDB family.

The gap between these two curves is rising steadily such that around 2013 there's almost a difference of 25%. One reason of this widening gap might be the selection of the data set: As I only restrict myself on the years from 1973 till 2013 (also with regard to the DocDB families), the chance increases over the years that patents, which were granted relatively late in that time frame, already have a DocDB equivalent from an earlier year in my data set.
However, that might not be the complete answer, especially if one takes a closer look on the DocDB family (DocDB family ID = 27290628) with the largest number of family members (195) inside the data set.

Now, what puzzles me, is that this family contains many granted patents, all of which have application authority US, application kind 'A', the same application filing date, even the same abstract, BUT different application IDs as well as application numbers.

Does anybody have a clue about the reason of this phenomenon?


Posts: 75
Joined: Thu Feb 22, 2007 5:33 pm

Re: Gap betweeen US patent counts & DocDB simple family counts

Post by EPO / PATSTAT Support » Mon Oct 29, 2018 9:46 am

Hello Thomas,
this is due to the endless series of (US) divisional applications originating from 6 "parent" US applications.
The process to assign DOCDB_FAMILY_IDs takes into account not only the so called Paris Convention priorities but also the data in TLS2016_APPLN_CONTN and TLS205_TECH_REL.

Have a look at this query that links all the DOCDB family members (docdb_family_id = 27290628
) with their parent applications :

Code: Select all

SELECT orig.appln_id 
      , orig.appln_auth 
      , orig.appln_nr 
      , orig.appln_kind 
      , orig.appln_filing_date 
      , orig.appln_filing_year 
      , orig.appln_nr_epodoc 
      , orig.appln_nr_original 
      , orig.ipr_type 
      , orig.internat_appln_id 
      , orig.int_phase 
      , orig.reg_phase 
      , orig.nat_phase 
      , orig.granted 
      , orig.docdb_family_id 
  FROM tls201_appln orig
  left join tls216_appln_contn on orig.appln_id = tls216_appln_contn.appln_id
  join tls201_appln parent on tls216_appln_contn.parent_appln_id= parent.appln_id
  where orig.docdb_family_id = 27290628
  order by parent.appln_nr_epodoc
PATSTAT Support Team
EPO - Vienna
patstat @

Posts: 2
Joined: Thu Oct 25, 2018 12:05 pm

Re: Gap betweeen US patent counts & DocDB simple family counts

Post by Tobor » Mon Oct 29, 2018 10:07 am

Thanks a lot for the answer :)

Post Reply