AmberBlog

Discussion of all things patent mapping and analytics.

  • Home
    Home This is where you can find all the blog posts throughout the site.
  • Categories
    Categories Displays a list of categories from this blog.
  • Tags
    Tags Displays a list of tags that have been used in the blog.
  • Login
    Login Login form

Can the 'licensing potential' of a patent be predicted?

Posted by on in Licensing Potential
  • Font size: Larger Smaller
  • Hits: 5497
  • Subscribe to this entry
  • Print

We have developed a metric that lets you predict the licensing potential of a patent. The benefit of this metric is that it is possible to efficiently review a large set of results to look for identify the most likely licensee. Below we demonstrate the value by example of a hybrid car case study, as well as a patent sucessfully asserted for more than a billion dollars in damages. 

The use of forward patent citations to predict potential licensing candidates is relatively well-known. Ambercite has also discussed this previously, including extending this to counting the number of forward citation patents owned by target companies, or identifying the most similar patents to the patents being licensed out.

But can these existing analysis techniques be further improved?

Logically, we could suggest a potential licensing candidate would have three desirable characteristics that would increase the potential for patent licensing:

  • An owner who may be in a position to infringe the patent. For example, if you own a patent covering a smartphone screen, Apple may be in a position to infringe this, but the University of Washington is less likely to be in this position
  • A patent owner that owns later patents that are similar to your patent
  • A patent owner that owns an important patent in the field. The inventions protected by the more important patents are more likely to be commercialised than the less important patents.

And how might you determine these things?

  • Working out the owners of forward citation patents is often straight-forward.
  • It is possible to determine patent similarity using the similarity filter available in AmberScope. Every patent link in our database is given a similarity score based on the number of overlapping patent citations with other patents (in simple terms).
  • Patent importance can be predicted by our AmberScore metric, which is normalised so that the average granted US patent has an AmberScore value of 1 . More commercially important patents tend to attract higher values. As an example of this, the famous Steve Jobs patent US7479949 now has an AmberScore value of 34, but in practice any AmberScore of above say 5 is pretty good. While patent rankings metrics such as AmberScore can make some people nervous, they can be best understood as means of identifying the patents that more likely to be important. This in turn can reduce the errors in patent analysis that can come from the alternative assumption that all patents have equal value, which should be self-evidently wrong to most people.

Lets assume, hypothetically, that we have a patent linked to 3 forward citation patents owned by companies that might be in a position to deploy this invention. We can show this situation with the diagram shown below:

AmberScore-illustration.jpg

 

So without knowing any further about these three patents A, B and C, which would be the best place to start an infringement analysis? I would suggest patent C, as there is a strong similarity to the client patent, combined with a high AmberScore value.

For purposes of ranking these patents, it is useful to have just one metric to rank on, rather than having to choose between either Similarity or AmberScore. For this reason we have developed a new metric for his ranking, which we call "Licensing Potential'. This metric is the geometric mean of Similarity and AmberScore.

In this case:

  • The Licensing Potential for forward citation patent 'A' = 1.5 (being the geometric mean of '2' and '1.2', which is the square root of 2 * 1.2)
  • The Licensing Potential for forward citation patent 'B' = 1.8 (being the geometric mean of '4' and '0.8')
  • The Licensing Potential for forwad citation patent 'C' = 6.4 (being the geometric mean of '7' and '6.0').

So patent C has the highest predicted "Licensing potential". This can be very useful to know if you are presented with a large group of patents for review and do not know where to start.

 

Applications for the 'Licensing Potential' metric

The benefit of this metric is that it is possible to efficiently review a large set of patents to look for licensing options. 

You can even apply this process in reverse, to determine your infringement risks (from backward citation patents) for your patent portfolio. In this case you would be looking for patents owned by potentially hostile owners that are similar to one of your high ranking patents. This can be used to help manage infringement risk, for example, or as part of a due deligence exercise when buying a patent portfolio or company that owns patent porfolios.

 

Example - Successfully asserted Paice hybrid car patent

Hybrid car drivetrain pioneer Paice's patent US5343970, covering a hybrid electric vehicle, has been well studied by Ambercite over the years. This patent (and other Paice patents) has been successfully asserted against both Toyota and Ford, and litigation from Paice is ongoing

We estimate that this patent currently has around 329 forward citations (unlike some other patent databases, we do not double count separate forward citations to a patent application and then its subsequent granted patent). If these forward citations are ranked by licensing potential, the top 20 patents are as follows.

Rank

Patent number

Licensing Potential

Listed owner

Patent title

Similarity

AmberScore

1

US6209672, (1999)

55.3

PAICE CORP

Hybrid vehicle

147

20.8

2

US6554088, (2001)

45.2

PAICE CORP

Hybrid vehicles

118

17.3

3

US6338391, (1999)

33.3

PAICE CORP

Hybrid vehicles incorporating turbochargers

108

10.2

4

US7597164, (2006)

22.8

PAICE CORP

Hybrid vehicles

75

6.9

5

US6116363, (1998)

22.1

FRANK TRANSPORTATION TECHNOLOG

Fuel consumption control for charge depletion hybrid electric vehicles

48

10.1

6

US5713425, (1996)

21.8

FORD

Parallel hybrid powertrain for an automotive vehicle

45

10.3

7

US6367570, (2000)

20.7

ELECTROMOTIVE INC

Hybrid electric vehicle with electric motor providing strategic power assist to load balance internal combustion engine

48

8.8

8

US5842534, (1997)

19.7

FRANK; ANDREW A.

Charge depletion control method and apparatus for hybrid powered vehicles

47

8.2

9

US5697466, (1993)

19.1

EQUOS RESEARCH KK

Hybrid vehicle

48

7.6

10

US5806617, (1996)

19.0

EQUOS RESEARCH KK

Hybrid vehicle

36

9.8

11

US5841201, (1997)

18.2

TOYOTA

Hybrid vehicle drive system having a drive mode using both engine and electric motor

37

8.9

12

US5845731, (1996)

17.1

CHRYSLER CORP

Hybrid motor vehicle

37

7.7

13

US5568023, (1994)

16.4

GRAYER; WILLIAM

Electric power train control

33

8

14

US6018198, (1998)

14.7

AISIN AW CO

Hybrid drive apparatus for vehicle

21

9.9

15

US5846155, (1996)

14.6

AISIN AW CO

Vehicular drive unit

26

8.2

16

US5558595, (1995)

14.2

GEN MOTORS CORP

One-mode, input-split, parallel, hybrid transmission

16

12.3

17

US8565969, (2010)

13.9

CLEAN EMISSIONS TECHNOLOGIES INC

Over the road/traction/cabin comfort retrofit

39

4.8

18

US5820172, (1997)

13.2

FORD

Method for controlling energy flow in a hybrid electric vehicle

25

6.9

19

US6098733, (1996)

12.7

TOYOTA

Hybrid drive system for motor vehicle

26

6.1

20

US5887670, (1997)

12.1

TOYOTA

Vehicle power transmitting system having devices for electrically and mechanically disconnecting power source and vehicle drive wheel upon selection of neutral state

32

4.5

 

In this table we have highlighted patents associated with Ford in blue, and patents associated with Toyota (and its known subsidiaries and affiliates) in pink.

This table shows that:

  • Many of the highest ranked patents are other Paice patents. This is no surprise, and can obviously be eliminated from consideration for licensing analysis.
  • There are lot of Toyota group patents toward the end of this table.
  • The patents with the highest Licensing Potential appear to be very similar to the Paice patent.
  • None of the AmberScore values for these forward citation patents were exceptionally high, but there were some very high similarity values.

Note that in this data we have also included AnberScore and similarity data, so that the origin of the Licensing Potential values becomes clearer.

The benefit of this analysis is that we can keep going for all 329 known forward citations. And if we do so, the final result will be something like this:

 

Paice-analysis_20141028-064632_1.jpg

This shows that:

  • Toyota (whom Paice asserted their patents against first) has the highest estimated total licensing potential from this patent, even if they had only 30 forward citation patents in this network
  • Ford was has the second highest licensing potential, but over 65 forward citation patents.

So in this case, the results of this Licensing Potential analysis matches what was seen in actual litigation.

 

Example - Carnegie Mellon 'billion dollar' patent

Again this has been extensively reported on earlier. To recap, Carnegie Mellon University successfully asserted two of its patents, US6201839 and US6438180 against Marvel Corporation and ended up with a judgement of $1.4 billion. If we take the later of these two patents (US6438180), the top 10 listing of forward citation patents ranked by Licensing Potential looks like this.

 

 

Patent number

Licensing Potential

Listed owner

Patent itle

AmberScore

Similarity

1

US6938196, (2002)

12.8

QUALCOMM INC

Node processors for use in parity check decoders

11.6

14

2

US7184486, (2000)

12.4

MARVELL INT LTD

LDPC encoder and decoder and method thereof

12.5

12

3

US6957375, (2004)

9.6

QUALCOMM INC

Method and apparatus for performing low-density parity-check (LDPC) code operations using a multi-level permutation

9.3

9

4

US6961888, (2003)

8.8

QUALCOMM INC

Methods and apparatus for encoding LDPC codes

9.6

8

5

US7000177, (2000)

8.6

MARVELL INT LTD

Parity check matrix and method of forming thereof

4.5

16

6

US6965652, (2000)

8.3

MARVELL INT LTD

Address generator for LDPC encoder and decoder and method thereof

4.3

15

7

US7133853, (2003)

7.7

QUALCOMM INC

METHODS AND APPARATUS FOR DECODING LDPC CODES

6.1

9

8

US7673223, (2005)

6.9

QUALCOMM INC

Node processors for use in parity check decoders

4

11

9

US7072417, (2000)

6.5

MARVELL INT LTD

LDPC encoder and method thereof

3.5

12

10

US8095854, (2007)

5.7

DVTG LICENSING

Method and system for generating low density parity check codes

2.8

11

 

This table suggests that both Qualcomm and Marvell are potential licensees for this patent. However Qualcomm appears to have some sort of working relationship with Carnegie Mellon University, so maybe these parties have worked this matter out between themselves.

If we look at a summary of all the 74 known forward citations to this patent, we can see that Qualcomm leads Marvel as the leading "Licensing Potential" for this patent. Nonetheless, Marvel is a clear candidate for licensing of this patent based on this analysis - which of course is what was seen in the actual litigation.

CarnegieAnalysis.jpg

 

Summary

This analysis has shown how determining the "Licensing potential" of the forward citations to a patent you are interested in can;

  • help objectively and systematically suggest the candidates (if commercialising similar items) most likely to require a license to your patent,
  • as well as suggest the most likely forward citation patents that may describe the products that may be infringing this patent.

 

How can our clients get access to 'Licensing Potential' data for their patents?

Ambercite plans to provide Licensing Potential values for individual patents along with:

  • More conventional forward citation data
  • 'Ghost' forward citation patents - which are later patents that may quite similar to your patent, but where this similarity has not yet been recognised by the applicants or patent examiners.
  • Regular (~monthly) updates of this data

This will form part of a soon to be released Patent Watch service. We expect that this product will be of strong interest to all owners of patents or patent applications, especially as pricing for this fully automated service will start from US$120 per year,

Considering all of the other costs involved in filing and maintaining patents, US$120 or less to receive a regular report on the potential commercial value of each patent in your portfolio should be an easy investment to justify. Discounts will be available for clients with significant portolios.

We are now taking requests for a strictly limited free beta trial for this service. If you are interested in this service, please send an email, including the patent numbers that you want to watch, to watch@ambercite.com. We will provide up to a maximum of 3 patents per client in this beta trial.

We can also provide Licensing Potential data for very large portfolio - for example, we have just completed an analysis of the entire US patent portfolio of Microsoft. Contact us for further information if you are interested in learning more.

 

11 November Update  - Application of Licensing Potential analysis to the US patent porfolios of Apple and Microsoft

Who might have the upper hand in cross-licensing? Read more at this newly published blog.

 

Rate this blog entry:
2


Mike first developed an interest in patent data when working as a research scientist, and deepened this interest when working as an IP manager which led to his role at Griffith Hack. Mike has published in the areas of chemical engineering, patent management, the value of patents and the use of patent data in in a wide range of publications and forums, including the international journals Les Nouvelles, and Managing Intellectual Property.