Hi all
I am interested in extracting "unique" 4-digit(first four) ipc codes from a certain set of docdb families. I see that there are different versions of ipc (starting 2006) in my version of patstat (sep2009).
What I wonder is if I keep unique IPCs in each family based on the entire set of IPCs in patstat would that lead to some overlaps? That's would the same ipc code in one version be presented as a different code in another version? If on the other hand I stick to ipc version January 2006, would I be finding some patent families with no ipc class at all?
Best regards,
Joe
ipc class versions patstat 2009
-
- Posts: 176
- Joined: Tue Oct 19, 2004 10:36 am
- Location: Vienna
Re: ipc class versions patstat 2009
Hello Joe,
Unless you take a "distinct" first 4 digits of IPC codes, there will be duplications within a family. In the latest PATSTAT versions we have eliminated duplications within an application (on full IPC symbol).
Yes it is possible that applications have different classifications over different PATSTAT versions. PATSTAT will always use the latest IPC version. Applications can ( and will be) reclassified between releases. PATSTAT users should never worry about the fact that in the past, a certain IPC code was used, which is not used any more. (deleted entries) You will of course need to check your queries for obsolete IPC (or ECLA) codes between versions. (PATSTAT users also should keep this in mind when comparing statistics from other sources. Sometimes statistics are based on IPC codes "as published".)
Example: Y01 (ICO) was used to tag nanotech patents; it has now been replaced by B82 (ECLA & IPC), and Y01 is not used any more. The results from the query below will be very much different between PATSTATAPRIL2009 and APRIL2011.
use patstatapril2011
go
SELECT Year(b.appln_filing_date),
COUNT(distinct(b.appln_id)) as april2011
FROM tls209_appln_ipc a join tls201_appln b on a.appln_id = b.appln_id
where ipc_class_symbol like 'B82%'
group by YEAR( b.appln_filing_date)
order by year (b.appln_filing_date) asc
Best regards
Geert Boedt
PATSTAT Promotion Team
patstat @epo.org
Unless you take a "distinct" first 4 digits of IPC codes, there will be duplications within a family. In the latest PATSTAT versions we have eliminated duplications within an application (on full IPC symbol).
Yes it is possible that applications have different classifications over different PATSTAT versions. PATSTAT will always use the latest IPC version. Applications can ( and will be) reclassified between releases. PATSTAT users should never worry about the fact that in the past, a certain IPC code was used, which is not used any more. (deleted entries) You will of course need to check your queries for obsolete IPC (or ECLA) codes between versions. (PATSTAT users also should keep this in mind when comparing statistics from other sources. Sometimes statistics are based on IPC codes "as published".)
Example: Y01 (ICO) was used to tag nanotech patents; it has now been replaced by B82 (ECLA & IPC), and Y01 is not used any more. The results from the query below will be very much different between PATSTATAPRIL2009 and APRIL2011.
use patstatapril2011
go
SELECT Year(b.appln_filing_date),
COUNT(distinct(b.appln_id)) as april2011
FROM tls209_appln_ipc a join tls201_appln b on a.appln_id = b.appln_id
where ipc_class_symbol like 'B82%'
group by YEAR( b.appln_filing_date)
order by year (b.appln_filing_date) asc
Best regards
Geert Boedt
PATSTAT Promotion Team
patstat @epo.org
Best regards,
Geert Boedt
PATSTAT support
Business Use of Patent Information
EPO Vienna
Geert Boedt
PATSTAT support
Business Use of Patent Information
EPO Vienna
-
- Posts: 140
- Joined: Wed Jul 08, 2009 5:51 pm
- Contact:
Re: ipc class versions patstat 2009
Don't forget ipc_class_level="A" otherwise you'll double count the IPC classes from the Core set. The distinct function fixes this, but the Core set may give a different unique set than the Advanced set so you still should use this criteria.
________________________________________
Nico Doranov
Data Manager
Daigu Academic Services & Data Stewardship
http://www.daigu.nl/
Nico Doranov
Data Manager
Daigu Academic Services & Data Stewardship
http://www.daigu.nl/
-
- Posts: 63
- Joined: Mon Nov 09, 2009 8:48 pm
- Location: Italy
- Contact:
Re: ipc class versions patstat 2009
Hi nico
from 201104 PATSTAT shows the Advanced symbols
so the field IPC_VALUE is used for Classification value (inventive or non-inventive) I,N
so distinction among Adv and core is valid only until 201009
GL
from 201104 PATSTAT shows the Advanced symbols
so the field IPC_VALUE is used for Classification value (inventive or non-inventive) I,N
so distinction among Adv and core is valid only until 201009
GL
Re: ipc class versions patstat 2009
Sorry for the late reply (I thought there were no responses), and thanks so much for your responses.
One question though. Where could one find 1. information about old ipc classes being replaced with new ones in Patstat 2011, and 2. the definitions of the 4-digit IPC classes. The IPC class information provided at WIPO is very exhaustive. While that information is useful if one is looking for a specific technology, that is not so if one, like me, is using an old IPC-industry concordance scheme like the one by Ulrich Schmoch et al. In the latter case one needs to know all changes made to earlier IPC classes.
Any thoughts on this?
Best regards,
joe
One question though. Where could one find 1. information about old ipc classes being replaced with new ones in Patstat 2011, and 2. the definitions of the 4-digit IPC classes. The IPC class information provided at WIPO is very exhaustive. While that information is useful if one is looking for a specific technology, that is not so if one, like me, is using an old IPC-industry concordance scheme like the one by Ulrich Schmoch et al. In the latter case one needs to know all changes made to earlier IPC classes.
Any thoughts on this?
Best regards,
joe
-
- Posts: 140
- Joined: Wed Jul 08, 2009 5:51 pm
- Contact:
Re: ipc class versions patstat 2009
I also missed the subsequent replies.
If I remember correctly, PATSTAT 2009 contains IPC classes for IPC version 2006-01-01 (also known as "IPC8"). PATSTAT 2011 (October) uses the same version. So no need for any concordance within PATSTAT.
But if you are using very old IPC codes then you will need to map these to IPC8. For this you will need a concordance table, or a description of the IPC Revision. Knowing WIPO it's going to be a big mess. I mean technically it's all in order, but they are not good when it comes to academic user-friendliness.
As for the definition of the IPC classes, again we depend on WIPO. Unfortunately they do not offer us something user friendly such as an Excel table... but they do have XML files hidden away in their download section and you can parse these to extract your own table. I have done this for IPC8 and I will post the Excel file online soon.
If I remember correctly, PATSTAT 2009 contains IPC classes for IPC version 2006-01-01 (also known as "IPC8"). PATSTAT 2011 (October) uses the same version. So no need for any concordance within PATSTAT.
But if you are using very old IPC codes then you will need to map these to IPC8. For this you will need a concordance table, or a description of the IPC Revision. Knowing WIPO it's going to be a big mess. I mean technically it's all in order, but they are not good when it comes to academic user-friendliness.
As for the definition of the IPC classes, again we depend on WIPO. Unfortunately they do not offer us something user friendly such as an Excel table... but they do have XML files hidden away in their download section and you can parse these to extract your own table. I have done this for IPC8 and I will post the Excel file online soon.
________________________________________
Nico Doranov
Data Manager
Daigu Academic Services & Data Stewardship
http://www.daigu.nl/
Nico Doranov
Data Manager
Daigu Academic Services & Data Stewardship
http://www.daigu.nl/
-
- Posts: 140
- Joined: Wed Jul 08, 2009 5:51 pm
- Contact:
Re: ipc class versions patstat 2009
See also http://forums.epo.org/epo-worldwide-pat ... c2074.html
That should solve your problem, if only the data were accessible.
That should solve your problem, if only the data were accessible.
________________________________________
Nico Doranov
Data Manager
Daigu Academic Services & Data Stewardship
http://www.daigu.nl/
Nico Doranov
Data Manager
Daigu Academic Services & Data Stewardship
http://www.daigu.nl/