PCT Regional Phase applications and problems in family assignation

All family creating algorithms rely on priorities similarity in order to build up groups of applications sharing all or some of priority applications.

Priorities are indicated on patent document from applicant

Under Article 4A(1) of the Paris Convention, ‘Any person who has duly filed an application for a patent … or his successor in title, shall enjoy, for the purpose of filing in other countries, a right of priority …’.  One of the underlying purposes of the priority right is to address the negative consequences of the principle of territoriality in the patent system, providing applicants with a 12 month period, following the filing of an initial application for a particular invention, within which to make subsequent applications in other territories of the world. (see as source: http://www.vennershipley.co.uk/show-news-id-310.html)

In order to build families following such criteria in patstat computations are based on table TLS204_APPLN_PRIOR linking applications ids among the application in exam and it’s priorities.

So any algorithm aimed to build up priority based families starting from TLS204 should fulfill three requisites:

1    1)      To consider all applications who are the same with the same application id (otherwise same priority document could be counted as a distinct in different applications)

2    2)     To consider all priority documents (otherwise we may have isolated applications cause they have no priority but in reality should belong to the same family)

3    3)      To have no bias due to national legislations or local practices creating differences into data.

For the latter point I cite as an example German Federal Patent Court on 28 October 2010 has recently published (11 W (pat) 14/09): the priority right should be transferred before the declaration of priority is filed for the subsequent application (which may be after the date of filing the subsequent application). (see as source: http://www.vennershipley.co.uk/show-news-id-310.html).

Nowadays the impact of such law has not been investigated yet in term of family creation bias.

For the first and second case two examples may explain well what happens:

CASE 1: missing priorities

APPLN_ID	APPLN AUTH	PUBLN NR
056198368	WO	2008077360
000006737	EP	2122995

Above table shows out 2 application ids who are equivalent since EP patent is regional phase of WO.

So by checking espacenet data (see hyperlink on publication numbers) we see WO patent is, as expected, priority of EP patent.

Unfortunately in TLS204 both applications have no priority so there is no linkage allowing to put them in the same family.

Oddly another equivalent has WO priority in TLS204

APPLN_ID	APPLN AUTH	PUBLN NR
273042219	DE	112006004209

So by building TLS204 based families we would have 2 distinct family the former including WO and DE patents, the latter with EP patent alone.

CASE 2: same application with distinct appln_id

APPLN_ID	APPLN AUTH	PUBLN NR
000002864	EP	1978807
054633065	WO	2007089571

Same as previous point, EP is regional phase of WO. In espacenet EP has WO application as priority (as expected) + 2 other priorities (US20060763253P 20060130; WO2006US32385 20060818) same as WO patent.

Let’s look into TLS204: situation looks different

APPLN_ID	PROGR	PRIOR_APPLN_ID	PRIOR APPLN AUTH	PRIOR APPLN NR	PRIOR APPLN DATE
000002864	1	900001124	US	76325306	30/01/2006
000002864	2	900001125	US	3258506	18/08/2006
054633065	1	54611861	US	2006032585	18/08/2006
054633065	2	900001124	US	76325306	30/01/2006

One priority is in common, another differs. WO application is not listed as EP priority. So different priority set. 

Even if at a deeper look we can guess that US 3258506 and US 2006032585 are the same application, only the appln_nr is written with a different spelling ( NNNNNYY vs YYàNNNNN where Y = appln year, N= application progressive number in the year).

About point 1 and 2 they can be crosschecked by comparing the consistency of priorities with the link between WIPO patent and it’s regional phase.

As a matter of fact table TLS201 includes the field INTERNATIONAL_APPLN_ID providing a way to verify if such a connection is also established in priorities table TLS204.

First we must check if PCT patent is listed as priority of the regional phase application

Select

  patstat.tls201_appln.APPLN_AUTH,

  patstat.tls201_appln.APPLN_ID,

  patstat.tls201_appln.INTERNAT_APPLN_ID

From

  patstat.tls201_appln Inner Join

  patstat.tls204_appln_prior On patstat.tls201_appln.APPLN_ID =

    patstat.tls204_appln_prior.APPLN_ID And

    patstat.tls201_appln.INTERNAT_APPLN_ID =

    patstat.tls204_appln_prior.PRIOR_APPLN_ID

Above query, who finds in TLS204 records where the PCT application is priority to the regional phase, returns 0 records. This means TLS204 do not list PCT application as priority of it’s regional phases.

Oddly the opposite analysis (check which WIPO applications have as priority it’s regional phase) returns some data: 398 patents in 201110 dataset) that is obviously a mistake see FI WO0074604 a year 2000 patents having a 2004 priority…

Select

  patstat.tls201_appln.APPLN_AUTH,

  patstat.tls201_appln.APPLN_ID,

  patstat.tls201_appln.INTERNAT_APPLN_ID

From

  patstat.tls201_appln Inner Join

  patstat.tls204_appln_prior On patstat.tls201_appln.INTERNAT_APPLN_ID =

    patstat.tls204_appln_prior.APPLN_ID And patstat.tls201_appln.APPLN_ID =

    patstat.tls204_appln_prior.PRIOR_APPLN_ID

Eventually we compare the result of an algorithm grouping in the same family patents who share same priorities or are one priority to the others (same as docdb family) and we see how many applications which are regional phase of a PCT application are in the same family of PCT patent.

appln auth	PCT applications	differrent family #	different family %
'AT'	127104	11428	8,99%
'AU'	667818	25912	3,88%
'BR'	129755	3409	2,63%
'CA'	348950	7768	2,23%
'CN'	76663	1598	2,08%
'DE'	541581	11281	2,08%
'DK'	50701	2444	4,82%
'EA'	15938	476	2,99%
'EP'	889972	19284	2,17%
'FI'	20591	641	3,11%
'GB'	28877	811	2,81%
'HK'	28061	799	2,85%
'HU'	17071	565	3,31%
'IL'	56832	2007	3,53%
'JP'	507623	9642	1,90%
'KR'	14322	791	5,52%
'MX'	96446	2577	2,67%
'NO'	61106	2016	3,30%
'NZ'	36315	1807	4,98%
'PL'	24612	880	3,58%
'SK'	12999	333	2,56%
'US'	577784	49988	8,65%
TOT	4401392	169340	3,85%

 

 Results are filtered for appln_auth with > 10.000 patents showing average error is 3.85% but some application authorities (like Us and AT) are above 8%.