Google is now recognised as one of the world’s leading companies, with one of the world’s leading brands and 53,000 employees. But not so many years ago, Google was just a couple of friends at Stanford University. One of them, Larry Page, filed a patent application on 10 January 1997 for what eventually became known as the ‘Page Rank patent’ , which is nerd humour for a patent filed by Larry Page for ranking the importance of web pages. (Check out the excellent description of this invention found here). This in turn became one of the key concepts that underpins the now ubiquitous Google search engine.
This patent, and 10 closely family members, were originally the property of Stanford University, and in 2007 were all assigned to the National Science Foundation. The most important of these is the first first, US6285999 for a Method for node ranking in a linked database. This has an AmberScore value of 53, or 53 times the average value for US patents, which may not surprise many. This high score partiy reflects the fact that this patent has been cited 716 times to date – it is clearly an important patent.
Ambercite has just released Cluster Searching, which can analyse both the direct and indirect patent landscape for an group of patents. Given the importance of the Google patents, we were curious to review the landscape of later and similar patents to the cluster of Page Rank patents. This was particularly so as this analysis is now very fast: running a report takes about 10 seconds, and only requires a list of the relevant patents
To make this analysis manageable, we restricted the results to the 1000 predicted-to-be-the-most similar patents. And in terms of analysis, we analyssed the patents we found using the concept of Licensing Potential, which combines the similarity and predicted importance of forward cititation patents to help predict the licensing potential to their owners.
And what did we learn?
Results – what are the most important and similar patents?
The most similar later patent is US5920859, filed just three weeks after the priority date of the Page Rank patents and how owned by Sungard Reference Data LLC . This claims a Hypertext document retrieval system and method, and is very similar in subject matter to the Page Rank patents.
The patent with the highest overall AmberScore value is US6029195, filed in December 1997 for a System for customized electronic identification of desirable objects. This has an AmberScore value of 76, and around 1195 forward citations to date.
The patent suggesting the highest Licensing Potential (from the Page Rank patents) is US6269361, filed in 1999 for a System and method for influencing a position on a search result list generated by a computer network search engine. This has a predicted Licensing Potential of 61.4. Originally filed by Goto.com, this is now owned by Yahoo.
Which companies have filed the most follow-on patents from the Page rank patents?
By computing the Total Licensing Potential for leading patent owners in the forward patent landscape from the Page Rank patents, we can see who has filed the most follow-on patents. Not suprisingly, Google heads this list – followed by Yahoo (also big in internet search), IBM, Microsoft, Xerox and Amazon.
Besides listing the total Licensing Potential, this figure also shows the split of direct and indirectly connected patents. As you can see, the majority of the Google patents in this most similar 1000 list have direct citations to the Page Rank patents. But for the other listed companies, only a minority of their patents have direct conections. But just because a connection is indirect does not mean that it is not relevant.
For example, the most similar (by our rankings) patent that is only indirectly connected is US6078916, filed in 1998 for a Method for organizing information and now owned by IAC Search & Media Inc. This covers an alternative means of indexing websites.
As a matter of interest, the Yahoo patent suggesting the highest LIcensing Potential is US6269361, which is discussed above.
What about prior art?
Using very similar principles, the most predicted to be similar prior art patent is US5754939, filed in 1995 by Fredrick Herz for a System for generation of user profiles for a system for customized electronic identification of desirable objects.
Perhaps of more interest to anybody who ever may want to invalidate this patent, the most predicted to be similar indirect (not yet cite) prior art patent is US5488725, filed in 1993 for a System of document representation retrieval by successive iterated probability sampling. And for those whom are interested, this is quite similar to the Page Rank algorithm, and my view potentially more similar than the US5754939 patent listed above – and yet was not cited by any of examiners for the Page Rank patents.
A review of the 1000 most similar pat
Want to run a similar analysis?