The power of the unknown citation - Case study based on Nintendo vs iLife. Updated Jan 2018.
January 16 2018
What is an unknown citation? And how can these help you?
The concept of an unknown citation is unique to Ambercite. We have defined this as a patent which is similar enough to a starting patent where maybe it should be cited as prior art * - but it has not been cited - hence the 'unknown' term
*(or as a forward citation if has a later publication date)
Ambercite uses an advanced AI algorithm to find these unknown citations, and we know from our clients that they can be very useful in helping to invalidate patents. This can also be used in IPR processes as they have not previously been cited by the examiner.
But are these citations helpful and accurate? The answer is that question is, in some cases, absolutely' - as will be discussed in the case study below.
(It should be noted that this has turned out to be one of our most popular blogs since its initial publication in September 2017. Since Sep 2017 there has been a big improvement in Ambercite, so this blog has been revised to reflect the current version of Ambercite).
Case study - Nintendo vs iLife
IP Watchdog has brought to our attention the case of Nintendo vs iLife, a case where the Nintendo managed to invalidate 5 out of 6 cases owned by ILIfe, but have been still been penalize for a $10 million dollar damages claim for the remaining infringement
Here at Ambercite, we develop software for finding prior art to patents, and so we wondered - did Ambercite Ai predict this invalidations?
In the IP Watchdog review, the first invalidated patent listed was US7479890, for a System and Method for Analyzing Activity of a Body. This patent claims a sensor for detecting if a person has not moved for a period of time.
This results in turn becomes a useful benchmark for Ambercite. But before we test Ambercite Ai, it could be worth considering what a semantic or class code search would have found.
What would have semantic and class code searching have found?
A review of both titles and abstracts for the iLife and Hitachi patents show that very different technical terminology was used. The abstract in the iLife patent uses technical words such as environment, sensor, accelerative, static and dynamic, while the abstract in the Hitachi patent uses different words such as motion, measurement, and actions. Neither set of terminology is wrong per se, but given this very different terminology it is very likely that semantic analysis would have struggled to suggest that these two patents were similar. And we do know that Boolean searching on the iLife patent failed to find the Hitachi patent.
And the same applies to class code searching. The iLife patent had 14 CPC codes and the prior art patent had 16 CPC codes - yet they only shared a single CPC code, being A61B5/1117, which covers fall detection. Fall detection is only part of the ILife or Hitachi patents, and in any case there are 1650 patents in the Espacenet database that claim this CPC code.
Given the above, it is no surprise that the examiner for the ILife patent did not find the Hitachi patent.
DID AMBERCITE AI PREDICT THE SIMILARITY OF THESE TWO PATENTS?
In contrast to semantic or class code searching, Ambercite AI finds similar patents by applying AI algorithms to our database of 150 million+ patent citation opinions. This is a very different approach that can find similar patents missed by other searching techniques.
To run an invalidation search in Ambercite can be as simple as entering the patent number in validity mode. In this case, because a European patent is listed as the invaliding document, we will preference EP family members# in the results. We have also set the date filter to only show patent families that were first published before the priority date of the iLife patent of September 1999.
The resulting query looks like this :
With the top 10 listed results (we can list up to 2000) found at this interactive link here.
Number 10 in the list is EP0816986 - which was the Hitachi patent used to invalidate the iLife patent.
Within Ambercite, there is also an ability to easily review these patents one by one, as shown below for the EP patent:
Of note, in this top ten list, nine out of the patents found are 'unknown' citations, i.e had not been previous cited against the iLife patent.,
# if the default setting of US patents had been used instead, the 3rd ranked patent would have been US6571193 - which is in the same patent family as EP0816986, but published later. Regardless, you would have ended up with EP0816986 as prior art for the iLife patent.
Does Ambercite Ai also work on known citations?
In an earlier blog, we showed how Ambercite could be used to find a known patent citation that was used to invalidate a patent. In this blog, we show the power of the unknown citation. In both cases, the only thing needed to run the analysis was the patent being challenged, and a few seconds of your time - our smart citation based AI algorithms took care of the rest of the search.
Do you want to invalidate patents yourself? Without the delays and costs of outsourcing patent searching?
Ambercite Ai is available on a subscription basis from Ambercite.
Please contact us to ask how you can join some of the world's leading patent owners, analysts, and patent attorneys, who use Ambercite Ai to strengthen their patent searching processes.
SPECIAL OFFER TO PATENT LITIGATORS - FREE PRIOR ART SEARCH REPORTS
Although Ambercite is in the business of selling patent search software and not patent searching, we are happy to provide a sample of our prior art search results to registered patent litigators. To take up this offer (one search per litigator) please contact us and let us know which patent would you would like a prior art search report for.
Why can we do this? Our product is very fast to use - and we are confident that you will like the results, and want to see more. Search reports are limited to one per litigator, and we will keep the results completely confidential.
Dependent on demand, we will aim to provide results within 24 hours, if not less.