Forward citations by country

Here you can post your opinions, ask questions and share experiences on the PATSTAT product line. Please always indicate the PATSTAT edition (e.g. 2015 Autumn Edition) and the database (e.g. PATSTAT Online, MySQL, MS SQL Server, ...) you are using.
Post Reply

UCBQ
Posts: 1
Joined: Tue Feb 28, 2023 6:28 pm

Forward citations by country

Post by UCBQ » Tue Feb 28, 2023 7:13 pm

Dear Patstat community,

I am interested in analysing the importance of national innovation systems for a particular technology field. For this, I would like to see the number of forward citations a country received over time. Ultimately, I am interested in seeing whether the respective inventions from a given country build a foundation for the specific technology and how this changed over time.

I’ve started out by using the following code:

Code: Select all

SELECT distinct  psn_name, person_ctry_code, appln_auth+appln_nr,  
appln_filing_date   ,nb_citing_docdb_fam
FROM   tls201_appln  
join tls203_appln_abstr on tls201_appln.appln_id = tls203_appln_abstr.appln_id
join tls207_pers_appln on tls201_appln.appln_id = tls207_pers_appln.appln_id
join tls206_person on tls207_pers_appln.person_id = tls206_person.person_id
join tls801_country on tls206_person.person_ctry_code = tls801_country.ctry_code
WHERE appln_filing_year between 1980 and 2021
and docdb_family_id  in (select docdb_family_id from tls201_appln where appln_auth = 'EP' and granted = 'Y')
and docdb_family_id  in (select docdb_family_id from tls201_appln where appln_auth = 'US' and granted = 'Y')
and docdb_family_id  in (select docdb_family_id from tls201_appln where appln_auth = 'JP' and granted = 'Y')
and appln_auth in ('EP','US', 'JP')
and granted = 'Y'
and applt_seq_nr > 0 and invt_seq_nr = 0
and (tls201_appln.appln_id in (select distinct appln_id from tls224_appln_cpc WHERE cpc_class_symbol like ('Y02E  10/728%')))
order by psn_name
I am using triadic patent counts (EP, US, JP) to limit the search to ‘high quality’ patents that have been registered in all three juristictions.
The CPC code is iterative for an industry, in this case onshore wind.

My idea is to now take the data and aggregate the nb_citing_docbd_fam by person_ctry_code and appln_filling_date to see how the number of citations has changed for different countries over time.

My first question is if this would make sense or if my approach is completely wrong?
I assume there will be a slight bias towards older patents although a rough examination of the data seems to suggest that if at all this would only be marginal.

Now to my second question: If the above approach makes sense, it should measure the number of forward citations each patent of the onshore wind class receives per country and year.
Would there be a way to do this kind of analysis for any cpc classification but limit citations to the onshore wind classification? This might not necessarily be that different but I would expect there might be technologies outside of wind energy that are being cited for wind patents?


EPO / PATSTAT Support
Posts: 425
Joined: Thu Feb 22, 2007 5:33 pm
Contact:

Re: Forward citations by country

Post by EPO / PATSTAT Support » Wed Mar 01, 2023 12:59 pm

Hello UCBQ,
Your approach needs some thinking over.
Using triadic patents is a good approach to select patents for which the applicant considers there is a need for a wider protection. Further restricting it to families where the triadic family members have to be granted patents reduces the set considerably. One has to keep in mind that it can take a couple of years before a patent becomes granted, and you will therefore have very few families where patents were filed the last 3-4 years, and already being granted. Maybe it would be good to be less restrictive on the condition that all 3 triadic members need to be granted,and simply consider that a triadic filing is sufficient.
Or remove the "grant" and only restrict to the fact that one family member is granted.--> "and docdb_family_id in (select distinct docdb_family_id from tls201_appln where granted = 'Y')"

You also have to see whether your research needs the data at patent application or patent family level. PATSTAT is mainly patent application based and not patent family (links to applicants, publications, citations, etc,...) If you want to aggregate at family level, then you will need to do this for all the data attributes you need for your research. Patent families can have different applicants in different countries, and in your set, you can observe that for quite some JP applications, the applicant data is missing. Country information for applicants having filed in JP is always missing. And therefore the JP family member will not be in your list because of the missing tls207_person_appln data.
You can observe this when you have a look at the results for the query looking at docdb_family_id = 34258238:

Code: Select all

SELECT tls201_appln.appln_auth+tls201_appln.appln_nr, appln_filing_date,tls207_pers_appln.*, 
tls206_person.person_name, tls206_person.psn_name,person_ctry_code, earliest_filing_id
  FROM 
  tls201_appln LEFT JOIN tls207_pers_appln on tls207_pers_appln.appln_id = tls201_appln.appln_id and applt_seq_nr > 0 and invt_seq_nr = 0
  LEFT JOIN tls206_person on tls207_pers_appln.person_id = tls206_person.person_id
   where tls201_appln.docdb_family_id = 34258238
The EP applicant is Senvion GmbH, the US applicant is REPOWER SYSTEMS AG and data for the JP applicant is missing. In fact, this company is now part of SIEMENS. (NULL values for JP are generated because we used a LEFT JOIN.)
To solve this, you could for example only use the applicant data from the EP application (which is nearly always there, and simply use the EP application as being the "representative application".
Something like this: (Observe that you still have duplicate rows at family level for those families that have more then 1 EP application.)

Code: Select all

SELECT distinct  psn_name, person_ctry_code, appln_auth+appln_nr,  
appln_filing_date   ,nb_citing_docdb_fam, docdb_family_id
FROM   tls201_appln  
join tls207_pers_appln on tls201_appln.appln_id = tls207_pers_appln.appln_id
join tls206_person on tls207_pers_appln.person_id = tls206_person.person_id
join tls801_country on tls206_person.person_ctry_code = tls801_country.ctry_code
WHERE 
 docdb_family_id  in (select docdb_family_id from tls201_appln where appln_auth = 'EP' and granted = 'Y'and appln_filing_year between 1980 and 2021)
and docdb_family_id  in (select docdb_family_id from tls201_appln where appln_auth = 'US' and granted = 'Y' and appln_filing_year between 1980 and 2021)
and docdb_family_id  in (select docdb_family_id from tls201_appln where appln_auth = 'JP' and granted = 'Y' and appln_filing_year between 1980 and 2021)
and appln_auth in ('EP')
and applt_seq_nr =1
and (tls201_appln.docdb_family_id in (select docdb_family_id from tls225_docdb_fam_cpc WHERE cpc_class_symbol like ('Y02E  10/728%')))
order by docdb_family_id,psn_name
Once you are there, you have to develop your methodology to select the forward citations because you specified that you want to have the forward citations at country level. You have to decide whether this is needed at the applicant or filing country level. You also specified you want to have the data "per year". This becomes rather complex as you can have multiple citations from the same family in different years.
For more in-dept knowledge on forward citations also see:
OECD patent statistics manual.pdf
(2.92 MiB) Downloaded 112 times
and And this:
spillover-evidence-in-forward-patent-ci ... 06#p39347
There are also plenty of posts going in depth on various issues of forward citations.
PATSTAT Support Team
EPO - Vienna
patstat @ epo.org


Post Reply