Disambiguation of inventors' names and addresses from patent data workshop

On May 30 will take place workshop on  “Disambiguation of inventors' names and addresses from patent data”, organized by "Academic Patenting in Europe" (APE‐INV), a project funded by the European Science Foundation.

The meeting will be hosted by the Department of Informatics, Systems and Communication (DISCO) of University Milano Bicocca.
It is meant to provide the possibility for computer scientists to meet lead users of inventors’ data from the fields of economics and management of technology, for an exchange of ideas over methodologies and tractability of disambiguation problems.

For any query, information or to confirm your presence, please contact dr. Monica Coffano (monica.coffano@unibocconi.it).

Program will be soon published @ this web page

How to get missing country code from homonyms in patstat applicants

In october 2010 patstat ediction we find in table TLS206 37.428.107 distinct person id (applicants or inventors); we would expect (or hope) they to have, a part from name, some geographic data, but as stated in some previous posts, a lot of them miss all informations a part from name making data quality improvement a little harder.
Exactly 13.032.871 persons (a 28% of the total) have no country code (and obviously in most cases no city, address etc.).

I start here a some posts about how to try to find clusters where it's possible to improve data quality of countries.

The first case I'll investigate applicants with no country codes.

let's take FI patent AP273 invented by BRUCE HOWARD DIXON [US] and applied by HOWARD DIXON BRUCE (with no country);
If we look at the PDF of the patent we see Bruce as applicant is listed as living in Florida, US, and as inventor is listed as "SEE ABOVE"; so we may presume a lot of applicants with no country when they have omonims in the same application, can inheritate country code from homonim inventor.

We must anyway remove possible doubles, like this couple of us patents, A & B where same applicant (and person id) Abate Riccardo invents a patent first as US, latter as IT.
So when creating correction table we must remove multiple occurences of same persons.

Applying this procedure in a simple way (I mean no standardization of names, just sheer string match) 472.668 (3,7% of missing)  persons with no country code can be assigned a country code.

Hidden non patent references

 This is no more than a note, born from a email exchange with Éric Archambault from Science Metrix.
 He was raising a question: do patstat count ALL non patent references listed in the patent?
 
Let's take as example patent USPTO  6365360 :
Table TLS214 lists 38 NPL references, those listed in the first pages of the patent.
It does not count the one 'hidden' in the body of the patent. We may see these 2:
 
 page 15:

These synthetic mRNAs are injected into Xenopus oocytes (stage 5-6) by standard procedures [Gurdon, J. B. and Wickens, M. D. Methods in Enzymol. 101: 370-386, (1983)]. Oocytes are harvested and analyzed for IP expression as described below.

Page 16:
and 9016 or other CMV promoter vectors, co-transfected with pDLAT-3 containing the thymidine kinase gene [Colbere and Garopin, F., Proc. Natl. Acad. Sci. 76: 3755 (1979)] in APRT and TK deficient L cells, selected in APRT (0.05 mM azaserine, 0.1 mM adenine, 4 ug/ml adenosine) and amplified with HAT (100 uM hypoxanthine, 0.4 uM aminopterin, 16 uM thymidine).

So, the answer is: NO,  in patstat table TLS214 we have only those listed in the 'reference cited' chapter...
this means that, if the examiner believes that such applicant's references are relevant, they will be listed, otherwise not.

ECLA coverage in Patstat

As previously stated, a new table (TLS217) has been recently changed in order to allow users to have ECLA (European classification system) and ICO data. (be aware in espacenet only ECLA will be indicated cause EPO application management decided that using the term ICO would only create confusion)

 I calculated a time series for coverage rate of ECLA for some selected application authorities (the coverage is never 100% due to applications D2 and unpublished priorities)

ctry	year	app#	ecla app#	rate
'AU'	1990	36251	30301	83,6%
'AU'	1991	34788	27117	77,9%
'AU'	1992	33847	27308	80,7%
'AU'	1993	35066	28061	80,0%
'AU'	1994	47278	34999	74,0%
'AU'	1995	51287	38489	75,0%
'AU'	1996	58047	45081	77,7%
'AU'	1997	65698	52721	80,2%
'AU'	1998	74505	59884	80,4%
'AU'	1999	82800	66252	80,0%
'AU'	2000	100118	82115	82,0%
'AU'	2001	108967	91214	83,7%
'AU'	2002	61498	47992	78,0%
'AU'	2003	108648	97395	89,6%
'AU'	2004	30908	22849	73,9%
'AU'	2005	32393	24428	75,4%
'AU'	2006	32508	24819	76,3%
'AU'	2007	31329	24315	77,6%
'AU'	2008	28592	23256	81,3%
'AU'	2009	8106	6756	83,3%
'AU'	2010	1707	1364	79,9%
'BR'	1990	10851	5385	49,6%
'BR'	1991	10122	4740	46,8%
'BR'	1992	9102	4683	51,5%
'BR'	1993	10272	5240	51,0%
'BR'	1994	10992	6160	56,0%
'BR'	1995	13554	7637	56,3%
'BR'	1996	15569	9701	62,3%
'BR'	1997	18562	11974	64,5%
'BR'	1998	19018	13194	69,4%
'BR'	1999	20987	14292	68,1%
'BR'	2000	20698	14020	67,7%
'BR'	2001	20588	13730	66,7%
'BR'	2002	19239	12435	64,6%
'BR'	2003	20878	13580	65,0%
'BR'	2004	22811	15196	66,6%
'BR'	2005	23922	16625	69,5%
'BR'	2006	13414	6627	49,4%
'BR'	2007	9197	2639	28,7%
'BR'	2008	7340	2201	30,0%
'BR'	2009	1712	1141	66,6%
'BR'	2010	11	2	18,2%
'CA'	1990	32986	28984	87,9%
'CA'	1991	31437	28248	89,9%
'CA'	1992	30629	28291	92,4%
'CA'	1993	28982	27070	93,4%
'CA'	1994	30365	27363	90,1%
'CA'	1995	30908	27382	88,6%
'CA'	1996	33304	29800	89,5%
'CA'	1997	36219	33416	92,3%
'CA'	1998	40014	37617	94,0%
'CA'	1999	41987	39584	94,3%
'CA'	2000	44197	41784	94,5%
'CA'	2001	43170	41082	95,2%
'CA'	2002	41726	39841	95,5%
'CA'	2003	42094	40245	95,6%
'CA'	2004	41836	39672	94,8%
'CA'	2005	43828	41964	95,7%
'CA'	2006	43777	42042	96,0%
'CA'	2007	40618	38841	95,6%
'CA'	2008	32917	31095	94,5%
'CA'	2009	7582	5679	74,9%
'CA'	2010	286	79	27,6%
'CN'	1990	32090	3993	12,4%
'CN'	1991	37717	3709	9,8%
'CN'	1992	48833	4368	8,9%
'CN'	1993	54913	7473	13,6%
'CN'	1994	57316	12462	21,7%
'CN'	1995	60234	15948	26,5%
'CN'	1996	67849	18995	28,0%
'CN'	1997	73307	22322	30,5%
'CN'	1998	79968	26654	33,3%
'CN'	1999	91154	28843	31,6%
'CN'	2000	112362	34774	30,9%
'CN'	2001	133048	41360	31,1%
'CN'	2002	165720	50178	30,3%
'CN'	2003	205557	62185	30,3%
'CN'	2004	235189	72010	30,6%
'CN'	2005	287662	81866	28,5%
'CN'	2006	341493	88081	25,8%
'CN'	2007	382948	86849	22,7%
'CN'	2008	404476	57082	14,1%
'CN'	2009	217326	15741	7,2%
'CN'	2010	168	9	5,4%
'DE'	1990	120133	109406	91,1%
'DE'	1991	119845	110519	92,2%
'DE'	1992	123828	113161	91,4%
'DE'	1993	127957	115675	90,4%
'DE'	1994	132197	119448	90,4%
'DE'	1995	135744	122253	90,1%
'DE'	1996	146343	131170	89,6%
'DE'	1997	154198	137603	89,2%
'DE'	1998	160431	142636	88,9%
'DE'	1999	164069	146394	89,2%
'DE'	2000	167977	152133	90,6%
'DE'	2001	160504	148403	92,5%
'DE'	2002	145041	134575	92,8%
'DE'	2003	134623	126893	94,3%
'DE'	2004	111554	103906	93,1%
'DE'	2005	105002	96516	91,9%
'DE'	2006	95404	89233	93,5%
'DE'	2007	83663	77257	92,3%
'DE'	2008	73819	61864	83,8%
'DE'	2009	25269	18843	74,6%
'DE'	2010	2800	894	31,9%
'EP'	1990	66699	66125	99,1%
'EP'	1991	62947	61982	98,5%
'EP'	1992	64930	63558	97,9%
'EP'	1993	65083	63563	97,7%
'EP'	1994	67463	65647	97,3%
'EP'	1995	71797	69406	96,7%
'EP'	1996	78950	76017	96,3%
'EP'	1997	89734	86132	96,0%
'EP'	1998	101724	97197	95,5%
'EP'	1999	111765	106112	94,9%
'EP'	2000	125923	119752	95,1%
'EP'	2001	134235	127466	95,0%
'EP'	2002	132311	124389	94,0%
'EP'	2003	137230	127106	92,6%
'EP'	2004	145312	135039	92,9%
'EP'	2005	154398	143605	93,0%
'EP'	2006	160288	149421	93,2%
'EP'	2007	160275	152546	95,2%
'EP'	2008	139610	130010	93,1%
'EP'	2009	54378	49541	91,1%
'EP'	2010	1537	1132	73,6%
'FR'	1990	15926	15858	99,6%
'FR'	1991	15857	15768	99,4%
'FR'	1992	15393	15316	99,5%
'FR'	1993	15505	15408	99,4%
'FR'	1994	15835	15725	99,3%
'FR'	1995	15955	15840	99,3%
'FR'	1996	16781	16592	98,9%
'FR'	1997	17782	17527	98,6%
'FR'	1998	18132	17842	98,4%
'FR'	1999	18586	18190	97,9%
'FR'	2000	19508	19095	97,9%
'FR'	2001	19654	19283	98,1%
'FR'	2002	19643	19208	97,8%
'FR'	2003	19308	18902	97,9%
'FR'	2004	19587	19032	97,2%
'FR'	2005	19777	19125	96,7%
'FR'	2006	19666	19111	97,2%
'FR'	2007	19399	18952	97,7%
'FR'	2008	19001	17569	92,5%
'FR'	2009	5571	4769	85,6%
'FR'	2010	166	95	57,2%
'GB'	1990	30053	14328	47,7%
'GB'	1991	30028	14125	47,0%
'GB'	1992	30290	14229	47,0%
'GB'	1993	30275	14332	47,3%
'GB'	1994	29560	14127	47,8%
'GB'	1995	29909	14785	49,4%
'GB'	1996	30448	15322	50,3%
'GB'	1997	31218	16147	51,7%
'GB'	1998	32828	16272	49,6%
'GB'	1999	35225	17287	49,1%
'GB'	2000	36994	17977	48,6%
'GB'	2001	36881	18722	50,8%
'GB'	2002	36315	18265	50,3%
'GB'	2003	35451	17954	50,6%
'GB'	2004	33788	17439	51,6%
'GB'	2005	30984	16693	53,9%
'GB'	2006	30029	15875	52,9%
'GB'	2007	30069	16144	53,7%
'GB'	2008	27638	14671	53,1%
'GB'	2009	25665	8049	31,4%
'GB'	2010	8461	724	8,6%
'JP'	1990	504479	71179	14,1%
'JP'	1991	482500	70490	14,6%
'JP'	1992	465127	70228	15,1%
'JP'	1993	443212	70985	16,0%
'JP'	1994	358575	71132	19,8%
'JP'	1995	375762	77305	20,6%
'JP'	1996	385899	84277	21,8%
'JP'	1997	398658	88601	22,2%
'JP'	1998	410097	89572	21,8%
'JP'	1999	414393	94508	22,8%
'JP'	2000	443598	103576	23,3%
'JP'	2001	448569	105439	23,5%
'JP'	2002	434335	107597	24,8%
'JP'	2003	432789	113556	26,2%
'JP'	2004	443034	121599	27,4%
'JP'	2005	447845	130070	29,0%
'JP'	2006	428966	133705	31,2%
'JP'	2007	405234	123480	30,5%
'JP'	2008	356748	80185	22,5%
'JP'	2009	66685	34792	52,2%
'JP'	2010	3361	884	26,3%
'US'	1990	132065	107893	81,7%
'US'	1991	138309	110782	80,1%
'US'	1992	145817	115837	79,4%
'US'	1993	152252	121461	79,8%
'US'	1994	167353	138544	82,8%
'US'	1995	192446	161648	84,0%
'US'	1996	194519	165358	85,0%
'US'	1997	226188	192533	85,1%
'US'	1998	235994	195596	82,9%
'US'	1999	267241	211016	79,0%
'US'	2000	320254	235039	73,4%
'US'	2001	374760	301396	80,4%
'US'	2002	384385	305251	79,4%
'US'	2003	400703	318661	79,5%
'US'	2004	450492	355179	78,8%
'US'	2005	490460	393708	80,3%
'US'	2006	478678	380253	79,4%
'US'	2007	473942	367942	77,6%
'US'	2008	423425	295259	69,7%
'US'	2009	223980	157793	70,4%
'US'	2010	17724	12686	71,6%

 We may note that data coverages changes with the time among different patent authorities (see FI Brasil or Japan in below chart), and that it has a decay in the last two years that can indicate a delay in inserting such info into patstat.