Patent examiners - Case study on using no-cost trial version of Ambercite AI to find hidden patents
Patent examiners and patent offices know that it can be very time intensive to run a patent examination search using traditional methods.
But even worse, after examiners might have spent hours and hours on a diligent search, there still can be relevant patents that they have missed, due to unexpected keywords or class codes - or in other words -' hidden patents'.
Ambercite has recently made a no-cost trial version of its Ambercite AI software available, and while this does come with some limitations, it can still be used to provide a rapid second opinion to a conventional search. This makes it an ideal complement to existing tools used by patent examiners and patent searchers (and other patent searchers)
Take, for example, the Nike patent WO2018004957, claiming a Digital golf assistant, and with a priority date of June 2016. This claims a app with the ability to set up virtual challenges.
Being a PCT patent it comes with an International Search Report, including 3 'Y' citations and 2 'A' citations.
But is this search complete? What can other similar but perhaps 'hidden' patents can be quickly uncovered using even the trial version of Ambercite?
How to find hidden patents using the trial version of Ambercite.
All we need to start with is the patent number - there is no need to predict keyword or class codes.
In the image below, we have set up a query on patent WO2018004957, asked for a validity search (this shows the earliest publication dates of the patent families found) and set the date filter so that it only returns patent families that were published (earliest family member) prior to July 2016. This query looks like this:
As expected, this returned 25 patents ranked in order of predicted similarity, and because this is a no-cost trial version, #1, 3, 5, 7 and 9 and subscriber only. But we can see #2, 4 and 6 in the partial screenshot below:
Of the remaining patents, 3 are known citations, and 18 are 'unknown' citations, i.e. not previously recognised as citations to the Nike patent family, but still predicted by our AI system to be relevant.
Because this trial version is available to anyone, there is no need to go through all of the results (you can easily do this yourself), but note the patent found in #16 position, namely US9344842, for a System and method for viewing golf using virtual reality.
This discloses, among other concepts, a series of virtual contests (which we could call 'challenges') that could be played against golfers, even if these other golfers are not present at the same time, for example:
The system and methods hereof allow for enhanced visualization of contests, easier logistics, and real time information. For example, a golfer might choose to participate in an “outing” when downloading the course information or “checking-in” with the course. Upon arrival at the contest for “closest to the pin” display 102 will show the golfer on the tee an AR message of the current leader of this contest, as well as the leader's distance from the cup or target.
How to quickly review results
Not only does this trial version of Ambercite give you this additional potentially relevant yet hidden patents in a simple manner, they are also quick to review. Selecting any of the pictures or abstracts brings up the patent review box, and we had requested that the keywords golf and challenge be highlighted in any field they exist in.
This same highlighting continues across to any of the Google patent link that we might select from the system - which you will note also includes a full set of pictures.
Using iterative searching to further expand the data set
Finding new prior art is helpful of course - and can even lead you to find further patents. For example, the patent US9344842 appears to be pretty relevant - so maybe we could use this as a starting patent to see what additional patents we can find?
This query is again simple to run - we simply replace the patent in the query box:
And re-run the search. However, being a follow up query, we can draw upon a special feature of Ambercite, in that new results are marked with a star:
By 'new' results, we refer to patents not previously listed in the current search cycle. So when we reviewing results, we can focus on these new results. And this soon brings us to US2007060408A1, not previously recognised as prior art to the Nike patent,
and which discloses:
A virtual golfer and club used can be rendered on the display at step 308 and a ball trajectory can be extrapolated at step 310 needed to arrive at a next stored location (such as a cub or other pre-stored location, e.g., from a prior stored game by the user or another participant).
Yet another hidden patent!
What additional value can the paid version of Ambercite provide to patent examiners and patent offices?
Besides allowing exporting of results, the paid version provides a lot more results, including the subscriber only patents in positions #1, 3, 5, 7 and 9. In fact the query in the full version can return up to 65 predicted to be similar patents, 43 of these patents being unknown patents both referring to 'golf' and to an aspect of IT. As an example of these other patents, US20140100007 discloses a System And Method For Measuring The Performance Of A Player In A Game, including:
wherein the game is comprised of a series of discrete trials wherein for each trial the participant receives a score selected from one of a finite set of discrete scores based on the participant's performance for each trial
A 'trial' in this context looks similar to the 'challenges' claimed by the Nike patents - and this patent was published as early as 2014. Yet this earlier patent was not cited against the Nike patent.
Different starting patents will return different numbers of similar patents. For example, if we will increase the patent number by one to WO2018004958, for Systems and methods for dynamic stitching of advertisements this would return a list of up to 796 similar patents. 796 patents is almost an impractically high number of patents to review of course, but a patent examiner could review the highest ranked patents, and then scan the rest for the presence of certain keywords.
The paid version of Ambercite is available to patent offices in a completely confidential manner, and is designed to interface with other patent search tools, and can even be accessed via an API - please contact us for further information.
How can patent examiners and patent offices find hidden patents using the trial version of Ambercite?
Simply register online at the link below, confirm the email address you use to register with, and start searching. There is no cost to start the trial, and no payment or credit card details are required - and you can use either your corporate or a personal email address, at least for the no-cost trial.