No items found.
< Back to Blog Page

Issues with Counting Citations - How Many Forward Citations does US8341981 Have?

Issues with Counting Citations - How Many Forward Citations does US8341981 Have?

This will be a multiple-choice question but first let me give you a hint, the ‘981 patent was published on January 1st of this year.

So which answer represents the number of forward citations for US8341981?

a. 22

b. 0

c. 7

d. All of the above

Some of you might be saying, “Come on now Tony, this is a trick question. You told us up front that this patent was published on January first of this year, how could it possibly have any forward citations yet? It just came out.” So the answer is b. or zero, right?

Personally, I would say the answer is d. but if you made me commit to an actual number I would say c. or seven. Let’s take a closer look at the history of this patent and its patent family to see if we can come up with a good answer.

The ‘981 patent is assigned to LG, it covers a washing machine and I chose it randomly simply as an example (the fact that my wife just bought an LG washing machine has nothing to do with it, I swear). In some ways it doesn’t even offer the best example of the ideas I would like to share but it works well enough to get the points across. It was filed for in the US on November 20th, 2008 but claims priority to four Korean patent applications, three of which are dated November 21st, 2007 and the fourth is dated April 30th, 2008. If you look at INPADOC patent family data it will show that there are 57 additional documents in the extended family including US20090146536 which is the US pre-grant application which eventually went on to grant as the ‘981 patent. There is coverage across many different countries including a WO application. When you look at this patent in Public PAIR you will find that there is no continuity data associated with it so while it claims priority to the Korean applications it doesn’t have a direct linkage to the WO document.

While I was being facetious with the mock dialog above about number of citations being zero I suppose technically this is true. Citations are based on the referencing of discrete documents so yes, as of January 29th, 2013, the ‘981 document has zero forward citations. Of course, this is not the full story because at the very least we need to account for the US pre-grant application which is directly connected to it. The ‘536 pre-grant application has seven forward citations associated with it. If we look out over the entire INPADOC patent family we find 22 forward citations overall including two to the before mentioned WO document and several to additional US pre-grant applications some of which have not gone on to grant yet.

Going back to the multiple-choice question, an argument can be made based on this data that d. or all the above is in fact the correct answer. We have seen that ‘981 in and of itself doesn’t have any forward citations. We also know that if we combine the pre-grant application with the granted patent we get seven forward citations for the set and for the entire collection of 58 INPADOC family members the number of forward citations is 22. It all depends on your perspective and explicitly documenting what you are actually counting.

In some previous posts, I discussed the difference between One Document Per Invention (ODPI) vs. One Documents per Family (ODPF) and how I felt that the ODPF approach was dramatically under representing the investment made by patent assignees in investing in a patent portfolio. I also looked at some individual countries and what impact this situation had in them. In a similar way it is important to recognize the impact of patent families on citation counts but I think we would again be misrepresenting the impact of discrete inventions within a family if we simply used a family forward citation count for all of them. That being the case, what should patent analysts use then and as users what should we be communicating to database and tool producers about how forward citations (and backward citations as well) should be shared.

Hopefully, we all agree that at the very least there should be a quick, automated method for associating the non-redundant forward and backward citations between a pre-grant application and its subsequent granted patent. While these are discrete documents and yes, there are often times differences between them, I would argue that they should be equivalent. They are, after all, the same application.

Even better, as far as I’m concerned, would be counting all the non-redundant citations between all of the documents in the same basic (as opposed to extended) family. If there is a WO and a series of EP documents along with two US documents for instance which all share the same priority application number and essentially all have the same set of claims (country specific modifications not withstanding) then all of these citations should be associated with one another.

Finally, while I don’t like the idea of grouping all of the citations together within an extended or INPADOC family it would be great to be able to do that as long as the overlapping documents between the family members could be removed so they don’t inflate the count.

While we are on the topic of things I can wish for it would also be swell if we could easily determine the number of self-cites from other cites in these same collections of counts. I won’t get into the Holy War over the importance or lack there of of self-citations in this post but it would be nifty if I could push a button to decide whether I wanted to see them or not.

This doesn’t apply only to the US by the way. While I used a US example the very same principles apply to European patents and any other country system where discrete documents are published at different times during the examination process. This didn’t always seem to be such a major issue but I began noticing the impact this was having in my citation analysis projects last year. A few years had gone by since pre-grant applications had started publishing and I noticed suddenly not accounting for the relationship between these documents was having a significant influence on what I was reporting. Staying with this topic, in an upcoming post I will take a look at exactly how big an issue this currently is with a random collection of US patents. I will also take a look at European patents and see if a similar trend exists.

To receive our newsletter on machine learning and artificial intelligence in the field of Intellectual Property from our sister website, enter your email below:

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
patinformatics logo
linkedin logotwitter logo

©Copyright 2021 Patinformatics