Page 1 of 1

PATSTAT and Espacenet consistency

Posted: Fri Jul 13, 2018 5:55 pm
by Edwardsj
Hi,

Previously I used the Advanced Search service on Espacenet and the search page only has "Publication Date." When I started to use PATSTAT online, I wonder how should I select the proper 'date' to get a similar output as the Espacenet results?

For example, if my Espacenet query is DB=EPODOC&ST=advanced&TI=&AB=nano*&PN=&AP=&PR=&PD=20170101-20171231, how should I write my PATSTAT query?

I wonder if the following query makes sense?

Code: Select all

select *
from tls201_appln a
left outer join tls202_appln_title t on a.appln_id=t.appln_id
left outer join tls203_appln_abstr ab on a.appln_id=ab.appln_id
where a.earliest_publn_date >= '2017-01-01' and a.earliest_publn_date < '2018-01-01' and
(t.appln_title like '%nano%' or ab.appln_abstract like '%nano%')
Thank you very much!

Re: PATSTAT and Espacenet consistency

Posted: Tue Jul 24, 2018 3:50 pm
by EPO / PATSTAT Support
The Espacenet field identifier PD stand for “publication date”, and its PATSTAT equivalent is the attribute PUBLN_DATE in table TLS211_PAT_PUBLN. So your query is almost perfect. It just needs a join with the publication table as well:

Code: Select all

select *
from tls201_appln a
left outer join tls202_appln_title t on a.appln_id=t.appln_id
left outer join tls203_appln_abstr ab on a.appln_id=ab.appln_id
join tls211_pat_publn p on a.appln_id = p.appln_id
where p.publn_date >= '2017-01-01' and p.publn_date < '2018-01-01' 
and (t.appln_title like '%nano%' or ab.appln_abstract like '%nano%')
Just do not expect to get exactly the same result in PATSTAT and in Espacenet. Every system slightly modifies the data to better fit their purpose. E. g. while Espacenet indexes each word of title and abstracts , PATSTAT treats title and abstracts as simple strings and is oblivious of any word or sentence structure.

Or Espacenet takes title / abstract from an English language family member, whereas PATSTAT has the title / abstract in original language (e. g. with Chinese characters).
Example: APPLN_NR_EPODOC = 'CN20172219392U' .

I hope this helps,
Martin