US Pre-Grant Applications have Significant Impact on Citations Associated with Granted Equivalents – Issues with Counting Citations Part 2

In a previous post we looked at the question of how many forward citations US8341981 has? We used this as an example of how the definition of patent equivalents and how they relate to the concept of a patent family can have an impact on what is represented by analysts when they discuss how often a particular patent document or invention is cited and by whom.  The post generated a significant amount of discussion on LinkedIn and the PIUG Blog. One of the items that most commentators agreed on was that the mechanisms for counting patent equivalents and family citations on Orbit.com by Questel was among the best of the systems surveyed on the ‘981 patent. At the end of the post I wondered if the issue of accounting for citations to pre-grant applications was a significant concern when looking at the forward citations of the corresponding granted patent or if these were relatively isolated instances.

To examine country specific issues related to this I started with the US since that was the country I looked at with the ‘981 example. The same pre-grant vs. grant situation holds true for other authorities as well, especially in Europe, so I will also look at that authority in a future post. The following search was conducted to look at the situation in the United States: find US granted patents and pre-grant applications with an application date between January 1st 2004 and February 14th (it is Valentine’s Day this week, after all) 2004 and a corresponding publication year between 2004 and 2010. This search generated 51,058 pre-grant applications, granted patents, reissues and other assorted patent document kinds. Since I am doing the analysis of these documents in Excel and working with spreadsheets with more than 50K rows can be tricky I choose an application date range which would retrieve about that number of documents. I used 2004 as the application year since I wanted the documents to be relatively recent but still early enough that both the granted patents and the pre-grant applications would have time to accumulate forward citations. It was for this reason that I also limited the publication year of the documents to between 2004 and 2010. While I can still talk about the impact of pre-grant applications on the corresponding granted patent for recently granted patents with no forward citations I thought it would be more interesting to look at granted patents which had a little time to generate citations of their own.

This sample size is less than 1% of all the granted patents in the US for the last ten years but it is still large enough to provide us with some insight. Since it is taken from a representative time slice and doesn’t have any additional biases built in, it should be reflective of what we would likely see if we worked with a larger collection or the entire data set.

Within this collection there were 19,883 granted US patents. Out of these, 15,197 or 76% of the documents had at least one forward citation associated with them. This, in and of itself was a surprising result to me. I did not expect that so many of the documents would have forward citations associated with them. I can’t speculate directly on this item in this post but it might be worth delving into further in a future one.

Out of the nearly 20K granted patents in this collection, 18,017 or 91% of the documents had a corresponding pre-grant application associated with them. This number makes sense since the majority of the US patents filed for since 2001 will have pre-grant applications associated with them. We will use the sub-set of granted patents from the period which have pre-grant applications associated with them to look at the forward citation occurrences.

Out of the 18K, 4,409 or 24% of the grants do not have any forward citations associated with them. This 24% matches the trend from the slightly larger group of all grants from this time period as we would expect. Looking at the 4,409 documents without forward citations associated with the granted patent we find that 2,633 of the pre-grant applications or 60% of them have forward citations associated with them. So by our first measure 60% of the grants, even when they don’t have any citations themselves have citations associated with the corresponding pre-grant application.

While looking at this sub-collection I also wondered if these grants had zero forward citations because they were predominantly newer documents (regardless of the limitation I put on the set that it had to have been published before 2010) so I looked at the distribution of the publication year for these granted patents and compared it to the publication year distribution for the entire collection of grants for the period. The chart below shows this comparison:

A comparison of publication year distribution of all granted patents in this collection versus the distribution for the granted patents which don’t have any forward citations

As can be seen, while the slope of the lines in the case of the total collection is steeper (and the subsequent decline bigger) than the case of grants with no forward citations the two sets follow the same pattern with regards to the distribution of their publication year. As such, there does not appear to be a bias with regards to when the patents granted when looking at this collection of data.

If we look at the 13,846 granted patents with at least one forward citation we find that only 2,395 or 17% of the corresponding pre-grant applications don’t have a forward citation associated with them. Said another way, in 83% of the cases where a granted patent from this time period has a forward citation and a corresponding pre-grant application there are forward citations associated with both documents.

If we remove from consideration the forward citations of the granted patents and only look at the forward citations associated with the pre-grant applications we find that of the 18,017 cases, 13,846 or 77% of these have at least one forward citation. This number correlates well with the 76% discovered with forward citations associated with all of the granted patents in this collection as mentioned earlier.

Looking at all of these values it is clear that forward citations to US patent documents are much more prevalent than this author originally expected. It is also clear that correlating forward citations with patent equivalents is imperative since even in the circumstances when the granted patent doesn’t have any forward citation associated with it 60% of the corresponding pre-grant applications do have citations associated with them. At least in the case of the United States, it is essential that patent equivalents be reconciled in order to get an accurate accounting of the citations associated with the invention.

Comments 4

  1. It gets even more complicated than that.

    If you want to know the total number of citations to the technical content of the invention, consider searching the whole extended DWPI family. In this case, although ‘981 is not yet in DPCI, it is in DWPI on STN. The entire extended family covers 9 DWPI records containing 52 documents covering 228 countries. (Some of those countries may be counted in more than one record).

    There are plenty of citations to other members of the simple and extended family. The DWPI record with ‘981 has 7 documents in the simple family. The US PGP in the family is the only one with citations in DPCI; it has three citations, all from US granted. If you add up all the citing patent counts for all members of the extended family, there are 6 citing patents from two different authorities (US and CN), and 6 Citing Patent WPI Accession Numbers. The latter is pretty close to the idea of invention-to-invention citations, rather than document-to-document citations. The 3 US members of the extended family are PGPs, and they are cited by two different US granteds.

    So, in answering the question “How many forward citations does patent X have?”, I would say the answer is definitely, “it depends”. Strictly speaking, in DWPI the ‘981 patent itself has no citations. However, the other members of the simple family which define the invention in a narrow sense have 3 citations (all to the PGP from US granteds), and if you look at the entire extended patent family (the broadest measure of the invention), there are 6 citations from 5 publications (4 US patents and 1 WO citation to a Chinese member of the extended family).

    Donald Walter, Ph.D.
    Solutions Consultant
    Thomson Reuters
    don.walter@thomsonreuters.com
    thomsonreuters.com

    1. Hello Don,

      Thank you for posting these details. This is very helpful.

      Can you speak to how users might reconcile these differences in the various Thomson Reuters products such as Innovation and the various online hosts such as STN?

      Thanks again, Tony

Leave a Reply

Your email address will not be published. Required fields are marked *