1. Introduction to Patent Families
To protect an invention in a specific country, an inventor must obtain a patent in that country. If an inventor wants protection in multiple countries, they must obtain a patent in each of those countries separately. When an inventor obtains multiple patents for the same invention in different countries around the world, that collection of patents is called a patent family.
Just as with ‘real’ families with their step-brothers, half-sisters, foster-parents and second-cousins-twice removed etc., relationships between patents can be complicated, and the lines determining where a patent family starts and stops can be blurry.
Two patent family definitions have been used in the patent information sector since the 1970’s:
1.1 The INPADOC (or often just referred to as ‘Extended’) patent family
INPADOC (‘International Patent Documentation’) is an international patent collection, produced and maintained by the European Patent Office (EPO).
The EPO definition of an INPADOC patent family is: “A collection of patent applications covering similar technical content.”
The technical content covered by the patent applications in the family is similar, but not necessarily the same, so an Extended patent family is a collection of patent documents that are considered to cover similar inventions. Members of an Extended patent family will have at least one priority claim to a priority document in common with at least one other member - either directly or indirectly.
A priority document is an earlier patent filing from which later patent filings can claim priority and inherit the filing date of that priority document as their priority date.
In the example below, Patents 1 and 3 are indirectly linked, because, while each of them claims priority to a different priority document (Priority docs A and B), Patent 2 claims priority to both priority documents. As a result, Patents 1, 2 and 3 are grouped into one Extended family.
1.2 DOCDB simple (or often just referred to as ‘Simple’) patent family
The DOCDB database is also produced and maintained by the EPO. It contains bibliographic details of worldwide patents, specifically priority claims of each patent. DOCDB simple families are based on the priority data in the DOCDB database.
The EPO definition of a simple family is: “A collection of patent applications covering the same technical content”.
Members of a Simple patent family will all have exactly the same priority claims. The technical content covered by the members of a simple patent family is considered to be identical, across the members of the simple patent family. It follows that a simple patent family is a collection of patent documents that are considered to cover a single invention. (Spoiler Alert – there are lots of examples when they do not cover a single invention).
E.g., in the example below, Patents 1 and 2 share identical priorities (Priority doc A) and are therefore in one simple family. Patents 3 and 4 also share identical priorities (Priority docs A and B) which are not the same as those of Patents 1 and 2 and are therefore in a second Simple family. All the patents also share direct or indirect links and so are in the same Extended family.
2. Challenges with patent data
All patent data originates as a human filled in form (even if electronically completed), and so human error is unavoidable. It is estimated that there is one typographical error per 5,000 characters written by an average person.
These simple typographical errors can cause erroneous links between two patents, and therefore join what should be two independent INPADOC families into one INPADOC family.
By way of a real-world example, consider: US 2015/0330109 A1 – “ADVANCED BIKE LOCKING SYSTEM AND METHOD” (“US ’109”)
US ’109 makes a priority claim to a US provisional application: 61/445,284. At some point someone entered 61/445,285 by mistake.
61/445,285 is actually a priority document for: US 2016/0234850 A1 – “SERVICE-BASED APPROACH TO CHANNEL SELECTION AND NETWORK CONFIGURATION FOR TELEVISION WHITE SPACE NETWORKS”.
The end result is that patents relating to a bike locking invention by a company called ‘Urban Wheely Inc’ appear in the same Extended family as patents relating to a telecommunications network related invention belonging to ‘Interdigital Patent Holdings Inc’.
This is clearly an error. Patently’s algorithms detect the erroneous link.
Fun fact: the example above was one of the first of many such examples we discovered while developing our family building rules, and as a result these errors are known to the team at Patently as “Urban Wheelies” after the company name ‘Urban Wheely Inc’.
3. Detection of errors in priority claims
The majority of erroneous links between two patents are caused by typographical errors and are handled by Patently’s family checking algorithms. But there are also invalid priority claims.
US 10,313,057 (“US ’057”) is an example – “ERROR DETECTION IN WIRELESS COMMUNICATIONS USING SECTIONAL REDUNDANCY CHECK INFORMATION”, filed by Qualcomm Inc.
US ’057 claims priority from two Qualcomm provisional applications: 62/346,291 with a filing date of 6 June 2016, and 62/344,031 with a filing date of 1 June 2016.
We know that, as outlined in 35 U.S.C 211.01(a), for a patent to be entitled to claim the benefit the priority date from an earlier filed US provisional application, along with a number of other stipulations: “… each prior-filed provisional application must have the same inventor or at least one joint inventor in common with the later-filed application...”.
In this case, US ’057 and US 62/344,031 do not share at least one common inventor. As a result, US ’057 does not have a legitimate priority claim to US 62/344,031. Patently determines that the earliest priority date available to US ‘057 is the filing date of US 62/346,291, 6 June 2016 – a five-day difference.
US 9,301,296 B2 (“US ’296”) is another example. US ’296 references several related earlier filed documents in the text of the patent specification, including PCT application PCT/US05/03518 and US Provisional applications 60/540,032 and 60/540,586.
These relationships are identified by the EPO as priority claims and added to the list of priority documents for US ’296. This in turn causes all other assets linked to one of these three documents to be classed as related by at least one common priority and therefore in the same INPADOC family.
When Patently’s advanced AI and Machine learning algorithms analyses US ’296 it detects a problem. You can see the phrase "relates to" in the below excerpt from US ’296, without specifying how US ’296 relates to PCT/US05/03518.
These types of references do not satisfy the criteria in another section of US patent law, namely 37 C.F.R 1.78(d)(2), which states: “…The reference also must identify the relationship of the applications, namely, whether the later-filed application is a continuation, divisional, or continuation-in-part of the prior-filed nonprovisional application, international application, or international design application…”
Patently determines the earliest priority date available to US ‘296 as the filing date of US 60/544,521, i.e. 13 February 2004 – a two-week difference.
A five-day or two-week difference in priority dates may seem trivial, but there may have been some relevant prior art published between those dates. When millions of dollars are dependent on the outcome of a court case, these details can be critical.
Fun fact: with over 4 million patents filed per year, two weeks equates to approximately 150,000 potential prior-art documents that could be incorrectly ignored.
Patently’s algorithms check the links between patents and priority documents, and, when it detects errors such as those explained in the examples above, it severs whichever link is determined to be invalid. This results in a reset of priority dates and reorganisation of extended families. As our data quality improves and the AI gets smarter, more and more of these problem links will be identified and severed.
4. Families in Patently
The family checking and building techniques used in Patently make it easy for users to find groups of patents where the technical content is the same, where it is similar and where it is different. We call this the Genetic® family scheme.
All Genetic families are displayed using our proprietary C-TreeTM visualisation which can be colour coded in various ways (here is it colour coded according to Status):
5. Patently Extended Families
Patently’s Extended families are similar to INPADOC Extended families but are not limited to patents sharing ‘official’ common priorities in the same way INPADOC families are. Our algorithms identify patents filed in different countries, for the same invention, and group them into the same Extended family.
US 2012/0309374 A1 and GB 2491392 A help explain how this works. Both cases were filed on the same day (2 June 2011), by the same inventors (Alistair Campbell and James Tagg), and both specifications have identical titles, abstracts, descriptions, claims and images. They clearly relate to the same invention.
Please note: this is a continuous work in progress and not all Extended families like these have been identified and linked yet.
6. Patently Extended Family references
Our Extended family references encode date information. Here’s how to understand them:
7. Genetic™ families
The EPO defines the DOCDB Simple family as “… a collection of patent documents that are considered to cover a single invention…”. According to the DOCDB definition, whether patents cover a single invention is determined by priority claims: patents with identical priority claims are considered to relate to a single invention and are thus in a single simple family.
Patently’s Genetic family definition similarly groups patents with identical priorities. But we take a further step: when identifying different inventions, we take account of metadata such as titles and inventors in addition to priority claims.
According to the Genetic family definition, each Genetic family includes a full set of priority claim links such that some patents can be shared between different Genetic families. The most common form of sharing is of a patent (or patents) at the root of a Genetic Family, as we will now explain by means of two examples.
In this first example, Patently detects separate inventions arising from (US) ‘478; as all members have identical priorities, they are grouped into the same DOCDB simple family; Patently separates these into two Genetic Families, A and B:
Similarly, in this second example, Patently detects two distinct inventions arising from the same three US provisionals . As all members have identical priorities, they are grouped into the same DOCDB simple family; Patently separates these into two Genetic Families, A and B.
You can always check these by visiting Espacenet and looking at the DOCDB simple family groupings; note that in the first example above, both US ‘299 and US ‘699 are found in the same DOCDB simple family but we can tell by the titles that they are distinct inventions:
And finally, we can't forget about the Continuation-in-Part (CIP); because CIPs can introduce new subject-matter we create new Genetic families clustered around each CIP.
8. Genetic™ family references
Genetic families use the Extended family reference scheme, and receive their own alphabetical code, beginning with A, through to Z and onward with double letter references from AA, AB etc. for the very largest of Extended family (which are referred to in Patently as “Hyperfamilies”).
9. Standard Essential Patent (SEP) families
One of Patently's core capabilities lies in its robust and accurate SEP data, making it an invaluable resource for analysing Standard Essential Patents.
An SEP is a patent that protects a technology deemed essential to comply with a specific industry standard, such as 4G or 5G. Companies declare their patents as essential to standards organisations, forming the basis of licensing negotiations under FRAND terms.
Individual patents that are declared essential to a standards setting body such as ETSI, are often referred to as the “basis of the declaration” and by extension “Basis patents.”
Patent families can be large and as explained above, may cover several different inventions. It’s not uncommon for one invention within a family to be declared essential while others are not. For SEP analysis, it’s critical to count the right type of family.
At Patently, we define an SEP family as a Genetic family containing at least one Basis patent. Therefore, Patently’s SEP families are intended to be consistent with the ETSI family definition. This approach counts Genetic Families containing declared patents and does not count Genetic families without declared patents. This provides the most accurate and meaningful SEP family counts for analysis.
For example, the single Extended family below contains three Genetic families, but only two of them contain patents that have been declared essential (see the ‘B’ badge indicating a ‘Basis patent’).
This is one extended family, three Genetic families (A, B and C each including coloured links identifying the family members) and two SEP families (Genetic families A and B):
This enables clear, reliable SEP assessments and supports informed decision-making for licensing, litigation, and portfolio management.
10. Final remarks
Just as people can move families by being adopted, married, divorced or remarried etc., so too, over time, can cases found in one Extended family or Genetic family move to another. Cases may move for any one of a number of reasons, such as a new Continuation-in-Part claiming priority from what were two separate families and so merging them into one, by data quality improvements, or by manual correction. As a result, it is worth noting not only family references but also the specific patent application number(s) of interest to avoid possible misidentification in future.
If you notice any issues with how specific cases are grouped into Genetic families within an extended family or simply have questions relating to Genetic families, please let us know by email: ask@patently.com