AmberBlog

CBS Interactive is being sued by Blue Spike Inc, a company based in Florida and who have started similar actions against a range of other companies. And in fact, a review of this matter on the US Patent Pair website suggest 39 different companies are being sued by Blue Spike, including NEC Corporation, Fujitsu, Nielsen, Yahoo, Google, and Shazam.

Blue Spike are alleging that CBS (and these other companies) are infringing several patents, the first listed patent being US7346472. This patent claims a method of monitoring 'signals' such as broadcasts, songs, videos, and potentially any media work, including when the signal is sent through the internet. It is claimed to be an advance over prior art methods that monitor similar signals using added digital watermarks, but where the added watermarks can get lost during broadcasting, or may need special software to read.

At the risk of oversimplfying a complex patent, claim 1 of this patent can be understood to be read as something along the following lines: 

1. Receiving a signal;
2. Creating an summary ('abstract') of this signal in some way, where the summary is based on the signal itself, and where the summary still retains a 'perceptual relationship' to the signal;
3. Storing the signal in a database;
4. Analysing a query signal in the same way, and comparing this to the stored summary in the database.

The 'perceptual relationship' term in this claim is likely to be important, and as defined in the body of the patent:

may either mathematically discernible or a result of market based needs....The purpose is to afford a more consistent means for classifying signals than proprietory related text-based approaches. A simply analogy is the way that in which a forensic investigator uses a sketch artist to assist in determining the identify of a human

Later on in the patent, the applicant refers to an embodiment including data reduction techniques that led to a:

significantly data-reduced that is not likely to reflect the human perceptual quality of the signal

So there appears to be no limitation that the relationship be human perceptual, i.e. being mathematically discernible is sufficient.

This Blue Spike patent has a filing year of 2000, and claims a series of priority dates, with the earliest claimed priority date being July 1996.

But is this patent novel and inventive? Because if it wasn't novel and inventive over the prior art, any claim by Blue Spike Inc against CBS and any other companies that it has alleged infringement of this patent against would disappear.

This is the sort of question that is ideally suited for our new patent search tool AmberScope, which searches for patents through citation relationships to known patents in the same field to find patents that may be disclosure similar inventions.

In this case study, we started by entering US7346472 into the AmberScope search box, producing the following image. This shows US7346472 as a dot in the centre of a network of other dots, each dot which represents either a patent that is a backward or forward citation to this central patent. There are also some 'ghost patents' shown, which are selected patents that are connected to either forward or backward citations, and which AmberScope suggests might be relevant to the patent in the centre of the network.

Note there are mostly backward citations, as indicated by the blue lines (showing backward citations). The green lines show forward citations.

In the above picture, the 'Title Box' of the BlueSpike patent is shown. The Title Box of other patents can be shown within AmberScope by moving a mouse over these other patents. Pressing the Details button found in each title box opens up either a Google Patents or Espacenet webpage providing further details of the patent.

One option to simplify this network is use the filing year Filter to remove all patents later than 1997, since 1997 was the priority year of the BlueSpike patent. But in this particular case, there are still plenty of patents to have to go through.

Another trick when looking at a crowded patent network is to use the 'Next button' (bottom left hand corner of the screen) to systematically work your way through every patent in the map - it need not take long. The patent Title Boxes are then displayed in the order of their AmberScore value, which is a measure of patent connectiveness that is shown in every patent box. For example, the patent box for the BlueSpike patent shows an AmberScore value of 2.9 (in the orange coloured triangle) - where 2.9 means that this patent is 2.9 times as connected as the average US granted patent filed in the last 20 years. 

By using the Next button in this way, we ended up opening up US5606609 to Scientific Atlanta, which was highly rated (AmberScore value of 30) and seems a little promising as it referred to 'an identifier for invoking the processing of the security information'. This is not quite what we were looking for, but a step in the right direction.

US5606609 has 411 forward and backward connections, and any one of these patents could be what we are looking for. So we 'refocused' the patent network onto this patent, using the 'more' button found in its patent box.

After the patent network has been refocussed this way, we can start exploring its connection in the same way as before, i.e. by filtering out patents filed later than 1997, and then using the Next button to systematically work our way from the highest ranked patent downwards. Although the diagram below looks a little crowded, in practice using the Next button makes navigating this patent network very manageable in practice.

This in turn leads us to US5259025 to Audio DigitalImaging, which discloses the concept of using biometric data such finger prints in a compressed form as an identifier.

Is biometric data a 'signal'?

We will leave this to the lawyers to argue this one. But they may not need to, because the real value in US5259025 is what happens when the network is refocused on this patent, and again the patents are year filtered and the Next button is used to work our way through the patent network.

This shows a couple of patents that are very close to what we are looking for, including

• US4230990 to Lert et al – creates a broadcast signature which is used to compare documents, and so appears to quite similar to the Blue Spike patent, and was filed in 1979.
• US4993068 to Motorola – use of a biometric characteristic in a compressed form as and identifier, and the use of this for verification. This patent was filed in 1989.

The Lert patent discloses an invention very similar to the Blue Spike patent by itself, but why stop when we are on a roll? So the network is refocussed one last time to be centered on US4230990, which has 377 further connections and an AmberScore value of 51, suggested that it is a very important patent. 

Using the Next button and the filing year quickly brings up two more patents of likely relevance to the Blue Spike patent:

• US4697209 to AC Nielsen– uses intrinsic parameters as a signature/abstract, and which are compared to other signatures, and was filed in 1984.
• US4843562 to  Broadcast Data Systems – waveforms are used as an 'spectagram' (similar to abstract) , and compared to other waveforms in a library, and was filed in 1987.

And we could keep going, but we have made our point by now. A number of these patents are a lot closer to the Blue Spike patent than the main patent discussed by the examiner in the file wrapper (US6088455 to Logan, published in 2000), and could potentially be of great interest to the lawyers defending companies such as CBS against charges of patent infringement.

This further confirms the ability of AmberScope, based purely on citation connections, to find highly relevant prior art missed by patent examiners and patent applicants.

While some of these patents are in different technical fields, they do disclose inventions that are similar in concept. And this case, we started with a patent that referring to sending signals, moved on to some patents that referred to biometric data, but then found these patents linked us to patents that were in a very similar technical field to what we started with.

Please note the overall searching techniques we need used in this search:

1. Start with the patent we trying to invalidate
2. Apply a filing year filters to make the search more manageable
3. Use the 'Next' button to work our way from the patent with the highest AmberScore value downwards, looking for patents of potential relevance.
4. Refocus the network on patents of potential interest, even if imperfect matches to what we are looking for
5. Repeat this process a number of times, working our towards patents of greater and greater relevance.
6. Do not worry about keywords or IPC codes, as these may be misleading and limit the patents that you find.

 This overall search strategy is shown below.

Interested in trying AmberScope for yourself? Simply go to www.amberscope.com and register your details to use AmberScope as part of our free beta trial. 

March 2013 update - some of the images shown above feature quite crowded patent landscapes. Thanks to an update in AmberScope introduced in February 2013, the same search would show a less crowded landscape which would be easier to navigate and faster to load - but still produce the same outcomes. 

 AmberScope is a new patent searching engine developed by Ambercite, in close cooperation with patent attorneys Griffith Hack, and able to find patents that may be missed by existing patent search processes. AmberScope is introduced in the video found here, but how well does AmberScope work in practice?

Earlier this year Google acquired three patents owned by Motion Research Technologies, and covering a pair of glasses for augmented vision. These patents included US7631968, filed in 2007, and which disclose a pair of glasses looking similar to this:

The claims of the patent cover the concept of the displayed image being controlled by the movement of the glasses. The granted patent includes a list of prior art documents, but did the examiner for this patent find all of the relevant prior art?

This is where AmberScope can be very helpful. The image below shows what happens if we run a search for this patent US7631968 ('968) in AmberScope.

Each dot represents a patent, with the dot surrounded by the red circle being '968, which was the 'focus' patent of this patent network. All of the patents connected to the '968 patent are shown, and these are represented in a light grey colour. Some other features are:

• Blue lines show backward citations, and green lines forward citations.
• The size of the dots represent its AmberScore rating. AmberScore is a proprietary algorithm that considers the influence of the patent in the network. In the image above, the '968 patent is shown with an AmberScore value of 0.24. This is lower than average – the average granted US patent over the last 20 years has an AmberScore value of 1.
• Thicker lines show stronger connections between two patents.
• The question mark in the patent box shows that the patent has not yet been rated by the user. AmberScope includes the facilities to capture 'personal ratings' on any patent, and this can be very handy for future referral. Currently, patents can be rated from 0 to 2.
• The image also shows some of the highly rated patents that are connected to the patents that are connected to the '968 patent. These indirectly connected patents could be regarded as influential 'friends of friends'. We refer to these patents as 'ghost' patents, and they are mostly shown as greyed out in this case. Ghost patents can easily be identified as patents with connection lines overlying them

     as opposed to the connection lines hidden behind the dots, which shows a directly connected patent  

Ghost patents can be very valuable, as these can disclose inventions that were not considered by the patent examiner (otherwise they would be listed in the search report) but still may be relevant.

But are these connected patents relevant to the '968 patent? 'Personal ratings' are just that, and so I went though and rated all of the connected patents, Figure 2. This also changed the colour of the dots, and the details of each patent rated are moved across to the table to the right after they have been rated.

In this figure, green patents are not thought to be relevant, blue patents are potentially relevant patents that are directly connected, and purple patents are potentially relevant ghost patents.

What do these patents disclose? The different patents all disclose different inventions, but one of the more interesting connected patents is US6349001, shown below, and which discloses: 

An eyeglass interface system is provided which integrates interface systems within eyewear. The system includes a display assembly and one or more audio and/or video assemblies mounted to an eyeglass frame

This is a relatively influential patent, with an AmberScore of 13, in other words 13 times as influential as the average granted US patent.

So already, AmberScope has shown its potential to find potentially relevant patents, even if in this case the examiner had also identified this patent as relevant prior art (hence the direct citation connection). But AmberScope includes another feature that can also assist in finding relevant prior art, namely the ability to 'walk the net', or refocus the patent network on any patent. This can be done as easily as selecting the 'more' button in the patent box as shown below. In this particular case, the button reads "178 more", which means that besides the connection between the '9001 patent and the '968 patent we started with, there are 178 other citation connections to this '9001 patent.

Selecting this 'more' button refocuses the patent network to be recentered on US6349001, allowing you to see it 178 direct connections, as well as its ghost patent connections.

Note that some of these patents are already coloured in yellow (the patent has been viewed), blue or purple This is because these patents were also seen in the network focused on the '968 patent, and AmberScope that remembered your previous ratings for these patents and transferred these ratings across. Practically, this means that you do not have to review these patents again, and instead can focus on the 'new' patents, which are coloured in grey.

But are any of these patents relevant to the Google glasses patent? To do this, we need to review the individual patent nodes - luckily this does not take that long within the AmberScope interface. As we do this, one of the more interesting patents we find is US5585871, which discloses:

A display apparatus secured to a temple or bridge contacting portion of an eyewear, the apparatus including means for monitoring the wearer's heart rate, lap position, laps completed, time elapsed, etc. An image of the collected data is transmitted into the wearer's field of view by means of a fiber optic element and projected at a focal point within the focusing range of the wearer's eyes.

This is not exactly the same as the Google glasses patent, but does include the element of augmented vision. Hence we have identified a second means of finding patents relevant to a starting patent.

Which is also potentially relevant.

But so are many of the 'ghost' patents that were shown when the network was focused on the  '968 patent, for example US6091546, which discloses

An eyeglass interface system is provided which integrates interface systems within eyewear. The system includes a display assembly and one or more audio and/or video assemblies mounted to an eyeglass frame. The display assembly is mounted to one temple and provides an image which can be viewed by the user

 This is an influential patent, with an AmberScore of 17 and 189 further connections. This is a third way of identifying relevant patents.

Ghost patents can also be used to refocus the network, which is what we have done below:

And when we do so, some of the new patents that we find when browsing this network may also be relevant to the '946 patent, for example US5719588, which discloses:

A viewing device for receiving video signals and generating corresponding images for viewing comprising a frame or support, adapted to be worn on the user's head, for example, a frame similar to a spectacle frame

So we have a fourth way of finding potentially relevant prior art, namely finding patents connected to ghost patents. And of course, we could make any of the patent in the new graph a focus patent, and continue to 'walk the net', and search for more relevant patents.

 Summary

In this short discussion, we have shown how it is possible to find potentially relevant prior art, some of it missed by the patent examiner, simply by starting with the patent number you are concerned about. This potentially relevant prior art could be

• directly connected patents (US6349001 in this example)
• patents connected to directly connected patents (US5585871)
• ghost patents (US6091546)
• patents connected to ghost patents ('friends of friends of friends', or  US5719588 in this example)

These different mechanisms are summarised in the diagram below.

And if you don't have a suitable starting patent, you could conduct a simple search for a start patent by running a simple conventional patent search for a starting patent which is close, but not close enough, to what you are looking for.

It is also worth considering what we have not done in this search:

• We have not looked at any keywords or semantic terms. Different patent applicants can use different keywords for the same concepts, and this can cause errors when searching for patents using keywords.
• We have not looked at any IPC or USPTO patent codes, which can be imprecise or incorrect
• We have not spent hours and hours looking long lists of patents, many of them irrelevant. Instead we have relied on the power of citation networks to quickly identify relevant prior art, some of which appear to have been missed by the original patent examiner.

And yet in this short demonstration we are only using part of the capability of AmberScope. Future blog posts will discuss these other capabilities and how they can assist you in finding relevant patents.

Postscript - comparison of keyword and patent codes in the patents found to the starting patent 

This is a good demonstration of how searching using keywords and patent codes could give you misleading results.

March 2013 update - some of the images shown above feature quite crowded patent landscapes. Thanks to an update in AmberScope introduced in February 2013, the same search would show a less crowded landscape which would be easier to navigate and faster to load - but still produce the same outcomes.