Monday, July 11, 2011

Testing application/publication link (PART II)

(Continues from previous post) 

As for linkage among more publiation numbers referring to the same application, now we check the reverse cases.

Double application by punr

On the other hand we may expect that the same publication number would refer always to the same application; this ‘habit’ cannot be applied to all patent authorities as we can see from table below where we calculate by patent office the share of publication numbers with more than one application id against the number of distinct publication numbers.

publn_auth
distinct punr
punr double appln_id
share
'AP'
7141
1
0,01%
'AR'
83260
6
0,01%
'AT'
514356
413133
80,32%
'AU'
1467157
506
0,03%
'BA'
198
22
11,11%
'BE'
583351
38
0,01%
'BG'
52604
2
0,00%
'BR'
466286
36436
7,81%
'CA'
1843352
350
0,02%
'CH'
704132
2323
0,33%
'CN'
3432787
303473
8,84%
'CO'
11558
1
0,01%
'CU'
2462
2
0,08%
'DD'
232684
779
0,33%
'DE'
5379549
626555
11,65%
'DK'
353081
30797
8,72%
'ES'
841106
72023
8,56%
'FI'
261990
13965
5,33%
'FR'
2311307
139
0,01%
'GB'
3090717
4
0,00%
'GR'
98672
2
0,00%
'HR'
11317
24
0,21%
'HU'
176105
6308
3,58%
'IE'
104206
130
0,12%
'IL'
166013
7520
4,53%
'IS'
8088
1726
21,34%
'IT'
825032
72099
8,74%
'JP'
13674415
4371804
31,97%
'KE'
1338
14
1,05%
'KR'
1866890
131794
7,06%
'LT'
6171
61
0,99%
'LU'
60745
14
0,02%
'LV'
4763
14
0,29%
'MC'
2657
99
3,73%
'MD'
4997
246
4,92%
'MX'
181726
139
0,08%
'NL'
581841
5321
0,91%
'NO'
280865
138
0,05%
'OA'
12747
152
1,19%
'PE'
2137
1
0,05%
'PL'
301135
18050
5,99%
'PT'
85717
33
0,04%
'RO'
64496
165
0,26%
'RU'
570155
4
0,00%
'SE'
633534
38
0,01%
'SG'
47181
2170
4,60%
'SI'
18962
1
0,01%
'SK'
31200
62
0,20%
'SM'
696
8
1,15%
'SU'
1182425
7
0,00%
'TR'
40373
6
0,01%
'TW'
294406
51212
17,40%
'UA'
47891
12
0,03%
'US'
10512836
19
0,00%
'UY'
6665
13
0,20%
'VN'
138
26
18,84%
'YU'
42390
1476
3,48%
'ZA'
238035
167
0,07%
'ZM'
2735
1
0,04%


If we look closer the case of Austria: publication number 18 is an emblemathic case (see also http://worldwide.espacenet.com/searchResults?NUM=AT18&DB=EPODOC&locale=en_EP&ST=number&compact=false ).
We can list minimum 2 distinct applications (appln_id 66192897 is an unpublished priority so we may drop it) published with kind U1 and T; Application 90809 published as AT18B can be questioned: espacenet lists it with no data and it’s publication date of 31/12/9999 is doubtful.

PAT_PUBLN_ID
 PUBLN_AUTH
 PUBLN_NR
 PUBLN_KIND
 APPLN_ID
 PUBLN_DATE
90809
 'AT'
 '           18'
 'B'
88381
 '9999-12-31'
426269
 'AT'
 '           18'
 'U1'
348623
 '1994-10-25'
73192897
 'AT'
 '           18'
 'U'
66192897
 '9999-12-31'
817187
 'AT'
 '           18'
 'T'
719612
 '1981-03-15'


Another example may be on CN; punr 1274133 refers to different applications
(see http://worldwide.espacenet.com/searchResults?NUM=CN1274133&DB=EPODOC&locale=en_EP&ST=number&compact=false)

from TLS211:

6603107, 'CN', '        1274133', 'A', 6185806, '2000-11-22', '', 0
7102574, 'CN', '        1274133', 'C', 6566474, '2006-09-06', '', 1

We may infer that those patent office having a share > 1% and a total figure > 2000 of duplication are authorities where more than one serial of publication number is used.

'AT'
514356
413133
80,32%
'BR'
466286
36436
7,81%
'CN'
3432787
303473
8,84%
'DE'
5379549
626555
11,65%
'DK'
353081
30797
8,72%
'ES'
841106
72023
8,56%
'FI'
261990
13965
5,33%
'HU'
176105
6308
3,58%
'IL'
166013
7520
4,53%
'IT'
825032
72099
8,74%
'JP'
13674415
4371804
31,97%
'KR'
1866890
131794
7,06%
'PL'
301135
18050
5,99%
'TW'
294406
51212
17,40%

2a) us example

Even if figure is small (19) it may be raising over time cause publication kind P (plants) has a serial number restarting from 1 that will in time grow and more and more get duplicated with patent of invention (A type) series.

'US'
10512836
19
0,00%

This is the example

PAT_PUBLN_ID
 PUBLN_AUTH
 PUBLN_NR
 PUBLN_KIND
 APPLN_ID
 PUBLN_DATE
 PUBLN_LG
62997521
 'US'
 '     20705'
 'P3'
49996628
 '2010-02-02'
 'EN'
70126104
 'US'
 '     20705'
 'A'
56493551
 '1938-04-26'
 ''

CHECK: PUBLN_NR US 20705 in your db counts 1 or 2 patents? [should be 2]

 2c) did we exclude all?

Eventually we have cases where more than one application id is assigned to the same publication, but it’s correct to drop all of them but one (‘cause 9999 or D2)

One example is EP122624, where appln_id 66112269 should be dropped ‘cause is an unpublished priority

PAT_PUBLN_ID
 PUBLN_AUTH
 PUBLN_NR
 PUBLN_KIND
 APPLN_ID
 PUBLN_DATE
20544092
 'EP'
 '   0122624'
 'B1'
17724798
 '1988-09-07'
20544093
 'EP'
 '   0122624'
 'A3'
17724798
 '1986-11-20'
20544094
 'EP'
 '   0122624'
 'A2'
17724798
 '1984-10-24'
73112269
 'EP'
 '   0122624'
 'A1'
66112269
 '9999-12-31'
 

Anyway this is one example of a general rule for EPO where one publication number correspond to only one application.

Check: see if EP122624 in you database correspond to one only application. [should be 1]
(or if you prefer check id any epo publication refer to more than one application).


No comments:

Post a Comment