Similarity, AmberScore and Licensing Potential? What does it all mean? And how does this help you?

Here at Ambercite, when using Cluster Searching, we like to talk about two calculated metrics:

  • "Similarity"
  • "AmberScore"

And sometimes "Licensing Potential".

One of the most common questions we get is 'what is the difference?' And 'how can this help me?'

 

What is Similarity?

Similarity is exactly that, the predicted similarity between two patents based on our citation analytics.

While similarity of course is a self-evident concept, imagine it being applied to houses. For example, somebody might use a similarity metric to predict that: 

 is similar, but not identical to

What is AmberScore?

AmberScore is the metric that leads to the most questions. Essentially we developed AmberScore because we recognised that when it comes to searching patents, not all patents are equal. Some patents are more important than others, and to treat all patents as being of equal value owould be misleading.

So we developed a metric we could use to rank patents.

A good number of analysts have concluded that the number of forward and backward citations are important predictors of patent value, and we agree.

But why? Any patent application costs money to draft and file. So imagine a 'widget X'.

Without knowing anything about the widget, it is possible to make a prediction of its commercial value based on the number of patent applications for similar widgets.  If there are lots of patent applications for similar widgets, this means that lots of applicants have spent money developing similar inventions, and then protecting these inventions with patent applications. They would have done this because they think the widgets can create a lot of value. In other words, applicants have voted with their wallet. 

And that is without knowing what the widget is - so this is a purely objective test that is not based on the subjective views of the reviewer.  The advantage of objective tests is that they are scalable. So it is possible to review thousands of patents quickly - and then maybe a skiller subjective reviewer can focus on the most likely to be important patents. 

If you were file a patent application for your improvement of widget X, it would over time attract some prior art citaiions  - and in turn later applicants would cite your patent. As its prior art and forward citations build up, so would its AmberScore value.

AmberScore values further increase if there are a large number of recent forward citations that cite the patent. Why? If lots of people are filing patents in the same area, it is clearly a 'hot' area - people are voting with their patent applications.

So that is the premise of AmberScore. In addition, we normalise it so that the average AmberScore for US patents is 1.0. Anything above 1 is above average.  And AmberScore values can get pretty high - for example US patent 7663607, filed by Apple for a Multipoint Touchscreen (now that is a type of widget that has attracted a lot of patent filings!) has an AmberScore value of 96.1.

In housing terms, a high AmberScore patent could look like this:

Big-house.jpg

I should note that it can take a few years for the necessary citations to appear to identify a high AmberScore patent, particularly for recent and not yet examined patent applications. So very recent patents may not have high AmberScore values - just like a half built house can be hard to value (what could be the value of the house below?).  But it does not take that long for the citations to build up.

Can you trust AmberScore?

Some people get nervous about using a proprietary metric such as AmberScore because they do not understand 100% how is calculated. But this is simply a ranking score, to help you focus your efforts.

As an analogy, every time that you run a Google search, you are running a search based on a proprietary metric. Google has to rank the websites listed on the page somehow right? And from what we understand, Google also use 'social network' based metrics, similar to ours.

Lets think about this a bit further:

  • People are happy to trust Google because it works - despite not knowing what algorithms are used to create the ranking metrics. 
  • Users of Google are sometimes happy to look through a number of links to find the website they are looking for. They do not expect the top ranked website to be exactly what they want.

Much the same applies to our ranking. It does work,  and can work very well, but sometimes the best patent for your needs is not our top ranked hit (the same applies to Similarity and Licensing Potential).

 

What is Licensing Potential?

Lets say you were interested in the monetization potential of a patent, or group of patents. We would say that you should look for later patents that were both:

  • similar to yours - which you could assess using our Similarity metric. 
  • important - which you could assses using our AmberScore metric

Once you had identifed such patents, you could review who owns them and if the owners could potentially benefit from acquiring or licensing your patents. 

Similarity and AmberScore are both positive metrics, so it is easy to combine them. We have done so, to create the metric "Licensing Potential" - which is calculated as the square root of Similarity * Amberscore. Licensing Potential values are provided for patent searchs in Cluster Searching where we apply a date filter to search for patents after a nominated date. 

And in real estate terms? It is hard to find an exact analogy, but consider the house shown to the left in the picture below:

The house on the right suggests that the area is gaining value, which is why the owner has invested in building this larger house. You would imagine that this larger house would in turn have increased the value of the house on the left.  

Personally, given a choice between the house above (on the left) and say the house below (on the left) which has a much less welcoming neighbor - I know which house I would prefer to own. 

a1sx2_Original1_Bad-neighbor.jpg

So in real estate, your neighbors matter. Much the same applies in patents, and this is what Licensing Potential attempts to capture.

 

What is the value of these metrics?

Ambercite has created these three metrics of SimilarityAmberScore and Licensing Potential because we believe that all patents are not equal. Some are more relevant to your needs and some are more important.

The alternative is to regard all patents connected by citation connections as being of equal relevance - and all patents being of equal importance. 

Which to me makes about as much sense as saying all houses are worth the same value...or in other words, no sense at all. This is why we have developed these metrics - in order to help you focus your valuable time when reviewing patents, and their forward and backward citations.