Wednesday, July 8, 2015

Patstat CPCs counts by patent office


Recently I had to try to understand how good is the coverage of patstat in terms of CPCs for some of the most relevant application authorities.
Obviously there is no benchmark data available, thus the main concept I used was a comparison of average number of Cpcs by application authority/year in order to display whether the baseline had some discontinuities.
I also dropped data where the number of applications was below 10k.

I display the results below.
A couple of short comments:
1) only few authorities seem to have discontinuities int their average number of CPCs: AT, BR and IT somehow.
2) average number of CPCs can differ much among patent offices (see FI values above 2 for MX, FR versus an average 0.2 for JP and 0.5 for CN); this seems to indicate difference in examination process, but values below 1 also indicate possible errors in data.







year
AT
AU
BR
CA
CN
DE
EP
ES
FR
GB
IT
JP
KR
MX
RU
SU
TW
UA
US
1990
1,5805
1,0222
1,2443
1,4408
0,319
0,6848
1,1223
1,5077
1,8886
0,9416
0,8238
0,1386
0,7823


0,115


0,7347
1991
1,8534
1,0777
1,2416
1,5371
0,2501
0,7081
1,1752
1,5672
1,9282
0,9396
0,7817
0,1468
0,6869


0,1497


0,7179
1992
2,0021
1,2314

1,634
0,2463
0,7133
1,1792
1,6628
1,968
0,9351
0,8025
0,1558
0,6441


0,1708


0,7021
1993
2,0424
1,2887
1,5505
1,7356
0,3428
0,7074
1,1998
1,7068
1,986
0,9856
0,8661
0,168
0,559

0,419



0,6926
1994
1,9649
1,0341
1,6268
1,6973
0,5032
0,6946
1,1794
1,7314
1,9676
0,9847
0,8669
0,2014
0,4896

0,5025



0,6595
1995
1,9811
1,0222
1,543
1,6666
0,5663
0,6806
1,1286
1,7256
1,9316
0,9925
0,8117
0,2011
0,4196

0,5805



0,5947
1996
1,9141
0,9888
1,5517
1,5929
0,5651
0,6548
1,0688
1,6665
1,876
0,9942
0,8342
0,2022
0,4373

0,5507



0,598
1997
1,9004
0,9449
1,4826
1,5193
0,5625
0,6324
0,9858
1,667
1,8377
0,9898
0,8288
0,2012
0,4574

0,584



0,5325
1998
1,8352
0,897
1,5285
1,4542
0,5681
0,6172
0,9136
1,6449
1,8431
0,9694
0,8305
0,1996
0,4092

0,568

1,0407

0,511
1999
1,7457
0,8469
1,455
1,4137
0,5214
0,6146
0,8547
1,6162
1,8554
0,9476
0,9134
0,2036
0,3617

0,5118

0,7697

0,461
2000
1,6225
0,7958
1,4826
1,3732
0,4663
0,6169
0,7965
1,6489
1,8563
0,9262
0,9536
0,1978
0,3027

0,5013

0,7841

0,3945
2001
1,656
0,7811
1,5078
1,3952
0,4516
0,6452
0,7667
1,6923
1,8714
0,9591
1,018
0,2014
0,3956

0,4893

0,7597

0,353
2002
1,7174
1,1174
1,5313
1,4065
0,4049
0,6962
0,7802
1,744
1,9083
0,9843
1,0718
0,2096
0,4178
2,2191
0,4933

0,7513

0,3497
2003
1,7458
0,8989
1,5068
1,4185
0,3647
0,736
0,7664
1,7436
1,9423
1,0343
1,0689
0,2152
0,3862
2,2894
0,514

0,5934
0,3853
0,3419
2004
1,7494
1,378
1,448
1,4239
0,3348
0,85
0,7282
1,7534
1,8975
1,1297
1,0841
0,2112
0,3122
2,2233
0,5591

0,537

0,3124
2005
1,7678
1,3685
1,4751
1,3627
0,2924
0,8675
0,6956
1,7406
1,9154
1,2638
1,1522
0,2116
0,3297
2,1751
0,6032

0,5286

0,2911
2006
1,8316
1,3601
1,4945
1,3074
0,2581
0,911
0,6806
1,8097
1,9235
1,3325
1,0348
0,2236
0,3352
2,3045
0,6156

0,5412
0,3992
0,2881
2007
2,0192
1,3467
1,475
1,287
0,2306
0,9558
0,6755
1,8217
1,9513
1,3541
1,0811
0,2324
0,3822
2,0726
0,6335

0,5729
0,3067
0,2803
2008
2,2159
1,4345
0,8578
1,3487
0,1984
0,9881
0,6757
1,8483
1,9885
1,3961
1,4127
0,2364
0,4006

0,6282

0,5591
0,2352
0,2864
2009
2,5456
1,4981

1,412
0,1637
1,0438
0,7061
1,8853
2,0335
1,3971
1,7341
0,2607
0,3819
2,0523
0,6939

0,5626
0,132
0,3034
2010

1,5006

1,4812
0,1395
1,0705
0,6947
1,8731
2,1195
1,449
1,8386
0,2692
0,388
2,1157
0,6558

0,5983
0,0535
0,2955
2011

1,4532

1,4864
0,1106
1,0637
0,6948
1,6421
2,1896
1,5088
1,8135
0,277
0,3975
2,1235
0,4225

0,5993
0,0403
0,2861
2012

1,2387

1,3143
0,0804
1,0109
0,6665

2,1202
1,4576
1,8017
0,2395
0,3442
2,0942
0,2363

0,5584
0,0367
0,2684
2013

0,5364

0,746
0,0437
0,7309
0,895

1,1866
0,7998

0,2894
0,2791
2,1296
0,0695

0,4253

0,2985

5 comments:

Anonymous said...

have figured out what the variables cpc_value and cpc_position mean?

I cannot find any info on whether these are meaningful (ideally, signal of relative importance of a given CPC class?)

Unknown said...

Brilliant! Thanks Gianluca. However, you say "instead", but in the data the majority have both.

I don't think it is always the case that F and I are perfectly correlated, either, so maybe we need to interpret them as different measures (rather than substitutes)? Just a thought. Thanks again. -Luis

GL said...

you're right, I amended the comment

GL said...

CPC_POSITION should be helpful for indentifying main class (where = "F" first) where patent authority has the concept
CPC_VALUE gives indication of the value of the classification i.e. is the class symbol relating to the invention or to aspects not related to the invention (but in the application). I=Invention A=Additional (Non-invention)

Patent search in India said...

Nice post, I bookmark your blog because I found very good information on your blog, Thanks for sharing more information
Trademark search and registration | Patent Registration in India

Post a Comment