In a previous post I described the One Document Per Invention (ODPI) method for creating patent corpuses for conducting statistics on patent collections. In the simplest method I proposed all granted US patents and pending applications were counted as discreet inventions and were carried through into the final analysis set regardless of how many of them might occur within a single INPADOC patent family.
This may seem jingoistic to some and suggest a bias toward including patent documents from the US to the detriment of other countries. This is not really the case and as I mentioned any country can be used as the primary country but the impact of using One Document Per Family (ODPF) is most profound for the United States and Japan based on the way their patent systems are set up regarding the use of Provisionals, Divisionals and Continuations.
To provide some evidence for this contention I thought I would do some research on the number of individual patent documents which appear in INPADOC families from a country other than the US. I could have chosen any number of important patent jurisdictions but in this post I will look at European patents.
The first thing to realize when comparing the US and European patent systems is that while Europe allows for Divisionals they do not allow for Continuation-in-Parts. Since this is the case the expectation right off the bat would be that there will be fewer European members contained within a single INPADOC patent family. Can we build a collection of European patents and take a look at whether this is the case in practice?
To try and accomplish this I started with a search for European granted patents with a publication year of either 2005 or 2006 that did not have the US listed as a priority country in their history. This produced a collection of 81,311 patents of which I used the most recent 60,000. In this case I used the 2005-2006 time frame since I wanted to go back a few years so a family could have been reasonably been expect to have developed without going so far back as to create a situation where international filing potentially could have been very different than it is today.
Looking at the 60,000 EP granted patents without a US priority country listing I then eliminated all documents that had a US granted patent or application within its INPADOC family. This was done to prevent a large US family from dragging in additional EP documents. Keeping in mind the purpose of this study is to see if multiple EP grants are typically found within a single INPADOC family. If families with US members were included this would be difficult to ascertain. This cut the collection from 60,000 documents down to 15,147.
So now that we have our collection, what do the number of documents in each family have to tell us? 9,255 of the cases only list two European document numbers, the pending applications and the corresponding granted patent (keeping in mind that the original collection was made up exclusively of granted European patents). In all of these cases there is only a single European application number represented in the family.
An additional 5,025 instances were found where three European documents were found within the INPADOC family. In this case, as well as some other instances that I will get to in a moment this could represent a situation where one than one European application number was present in the family. To disambiguate this situation I performed the following steps:
In the case where there are three European documents present in a family only 138 of the 5,025 instances contained multiple, unique application numbers. So far these examples follow the expected course where in the case of two European documents present there is an A1 and a grant present while in the vast majority of the cases where there are three documents present they represent an A2, an A3 and the grant.
Again, this is to be expected but what happens when the number of European entries within a family starts to climb? Once we get to four we start to see the emergence of multiple unique application numbers within a single family. Using the same steps as we followed for the instance of three documents we find that out of 438 cases where there were four documents in the family, 204 examples were found with multiple application numbers. 201 of these cases involved two unique numbers while three of them involved three unique numbers.
Finally, there were 421 examples where there were five or more European documents within a family. All but three of these cases had multiple application numbers present as one would expect when the number of documents within an INPADOC family start to rise.
Adding up all of the example of multiple European family members within a single INPADOC family we find that 760 out of the 15,147 cases exhibited this behavior. By this method it appears that about 5% of the cases where there are no US family members and the US is not a priority country results in multiple European patents in the corresponding INPADOC family.
It is up to the analyst to decide if it is worth trying to incorporate each of the individual granted EP documents when conducting a ODPI analysis or if they can get away with ODPF when dealing with EP documents. From this one specific analysis it appears that ODPF is reasonably accurate when dealing with European patents with the majority of the families being represented by a single application number.