Harvard Dataverse
Revision as of 14:24, 29 July 2016 by MarcelaInteriano (talk | contribs)
This page records how to load/use the Harvard Dataverse. The patents from 1975-2010 loaded as .sqlite3 and csv files can be found at
Unlike USPTO raw data from 1975-2010, this is cleaned data. In particular, the Harvard Dataverse datasets are post author disambiguation. For details, see the paper at Harvard Dataverse. All of them have been downloaded to the database server and can be found at
cd /bulk/patent
Getting the data
Where we get the Harvard Dataverse data: Harvard Dataverse
For more information about the patent data: Patent Data
To recreate the tables:
1. run createtables.sql
2. run copytables.sql
3. run cleaning db.sql
These scripts are available under /bulk/Software/Database\ Scripts
Loading the tables into the database
COPY classes FROM '/tmp/class.csv' DELIMITER ',' NULL CSV; COPY assignees FROM '/tmp/assignee.csv' DELIMITER ',' NULL CSV; COPY citations FROM '/tmp/citation.csv' DELIMITER ',' NULL CSV; COPY inventors FROM '/tmp/inventor.csv' DELIMITER ',' NULL CSV; COPY patents FROM '/tmp/patent.csv' DELIMITER ',' NULL CSV; COPY patdescs FROM '/tmp/patdesc.csv' DELIMITER ',' NULL CSV; COPY lawyers FROM '/tmp/lawyer.csv' DELIMITER ',' NULL CSV; COPY scirefs FROM '/tmp/sciref.csv' DELIMITER ',' NULL CSV; COPY usreldocs FROM '/tmp/usreldoc.csv' DELIMITER ',' NULL CSV; COPY invpats FROM '/tmp/invpat.csv' DELIMITER ',' NULL CSV;
Accessing the dbase
On the database server, the entire Harvard Dataverse for Patent Data is downloaded in directory
/bulk/patent
The files have been processed to be used for the create table scripts I have created to load the Harvard Dataverse database into our own psql database.
These preprocessed files are available under
/bulk/patent/preprocessed
The database I have created for the Patent Data from 1975-2010 is called patent data. After logging in to researcher@128.42.44.181, you can get access to the database by
psql patentdata
You can see a list of the tables inside the patent database by
patentdata=# \dt
List of relations
Schema | Name | Type | Owner
--------+-----------+-------+------------
public | assignees | table | researcher
public | citations | table | researcher
public | classes | table | researcher
public | inventors | table | researcher
public | invpat | table | researcher
public | lawyers | table | researcher
public | patdesc | table | researcher
public | patents | table | researcher
public | scirefs | table | researcher
public | usreldocs | table | researcher
(10 rows)
You can select from the database, for example, by
patentdata=# SELECT * FROM CLASSES WHERE patent='03930270'; patent | prim | class | subclass ----------+------+-------+---------- 03930270 | 1 | 360 | 130.24 03930270 | 0 | 360 | 84 03930270 | 0 | 360 | 271.8 03930270 | 0 | G9B | 15.08 (4 rows)
Or, do a join:
patentdata=# SELECT * FROM inventors AS i, classes AS c WHERE i.Patent=c.Patent AND i.Patent='03930270';
patent | firstname | lastname | street | city | state | country | zipcode | nationality | invseq | patent | prim | class | subclass ----------+-----------+----------+--------+-----------+-------+---------+---------+-------------+--------+----------+------+-------+---------- 03930270 | YOSHINORI | YAMAZOE | | OSAKA | | JP | | | 3 | 03930270 | 1 | 360 | 130.24 03930270 | YOSHINORI | YAMAZOE | | OSAKA | | JP | | | 3 | 03930270 | 0 | 360 | 84 03930270 | YOSHINORI | YAMAZOE | | OSAKA | | JP | | | 3 | 03930270 | 0 | 360 | 271.8 03930270 | YOSHINORI | YAMAZOE | | OSAKA | | JP | | | 3 | 03930270 | 0 | G9B | 15.08 03930270 | CHIHIRO | NAKAMURA | | MORIGUCHI | | JP | | | 4 | 03930270 | 1 | 360 | 130.24 03930270 | CHIHIRO | NAKAMURA | | MORIGUCHI | | JP | | | 4 | 03930270 | 0 | 360 | 84 03930270 | CHIHIRO | NAKAMURA | | MORIGUCHI | | JP | | | 4 | 03930270 | 0 | 360 | 271.8 03930270 | CHIHIRO | NAKAMURA | | MORIGUCHI | | JP | | | 4 | 03930270 | 0 | G9B | 15.08 03930270 | SATOSHI | KIKUYA | | KADOMA | | JP | | | 2 | 03930270 | 1 | 360 | 130.24 03930270 | SATOSHI | KIKUYA | | KADOMA | | JP | | | 2 | 03930270 | 0 | 360 | 84 03930270 | SATOSHI | KIKUYA | | KADOMA | | JP | | | 2 | 03930270 | 0 | 360 | 271.8 03930270 | SATOSHI | KIKUYA | | KADOMA | | JP | | | 2 | 03930270 | 0 | G9B | 15.08 03930270 | SHIZUHIKO | TANIGAWA | | MORIGUCHI | | JP | | | 1 | 03930270 | 1 | 360 | 130.24 03930270 | SHIZUHIKO | TANIGAWA | | MORIGUCHI | | JP | | | 1 | 03930270 | 0 | 360 | 84 03930270 | SHIZUHIKO | TANIGAWA | | MORIGUCHI | | JP | | | 1 | 03930270 | 0 | 360 | 271.8 03930270 | SHIZUHIKO | TANIGAWA | | MORIGUCHI | | JP | | | 1 | 03930270 | 0 | G9B | 15.08 (16 rows)
The schema I used for this database is documented at the bottom of this page.
And this is also saved as a .sql script under
/bulk/Software/Database\ Scripts/createtables.sql
A corresponding Drop table scripts is also available at
/bulk/Software/Database\ Scripts/droptables.sql
Converting DataTypes
Method 1: Explicit recast
CAST function
Method 2: implicit recast
SELECT varname::type
Method 3: through a language
CREATE OR REPLACE FUNCTION chartoint (text) RETURNS int AS $$
if ($_[0]) {
if ($_[0] =~ m/^\d+$/) {
return $_[0];
} else {
return -1;
}
}
return undef;
$$ LANGUAGE plperl;
The scripts for cleaning up the tables are available under /Database\ Scripts/cleaning\ db.sql
Creating the tables
The following is the schema for the 10 tables in this database:
I have cleaned up and fixed the types for the the following tables:
CREATE TABLE patents (
Patent integer,
Kind varchar, #only types NULL, A1, B1, B2
Claims integer,
AppType integer, #int (http://www.uspto.gov/web/offices/ac/ido/oeip/taf/filingyr.htm) 29 is design
AppNum integer,
GDate date,
GYear integer,
AppDate date,
AppYear integer
);
CREATE TABLE classes (
Patent integer
Prim integer,
Class varchar,
Subclass varchar
);
CREATE TABLE assignees (
Patent integer,
AsgType integer,
Assignee varchar,
City varchar,
State varchar,
Country varchar,
Nationality varchar,
Residence varchar,
AsgSeq integer
);
CREATE TABLE citations (
Patent integer,
Cit_Date date,
Cit_Name varchar,
Cit_Kind varchar,
Cit_Country varchar,
Citation integer,
Category varchar,
CitSeq integer
);
CREATE TABLE inventors (
Patent integer,
Firstname varchar,
Lastname varchar,
Street varchar,
City varchar,
State varchar,
Country varchar,
Zipcode varchar, #??
Nationality varchar,
InvSeq integer
);
CREATE TABLE patdesc (
Patent integer,
Abstract varchar,
Title varchar
);
CREATE TABLE lawyers (
Patent integer,
Firstname varchar,
Lastname varchar,
LawCountry varchar,
OrgName varchar,
LawSeq integer
);
CREATE TABLE scirefs (
Patent integer,
Descrip varchar,
CitSeq integer
);
CREATE TABLE usreldocs (
Patent integer,
DocType varchar,
OrderSeq integer,
Country varchar,
RelPatent varchar,
Kind varchar,
RelDate date,
Status varchar
);
CREATE TABLE invpats (
Firstname varchar,
Lastname varchar,
Street varchar,
City varchar,
State varchar,
Country varchar,
Zipcode integer,
Lat double precision,
Lon double precision,
InvSeq integer,
Patent integer,
GYear integer,
AppYear integer,
AppDate date,
Assignee varchar,
AsgNum varchar,
Class varchar,
InvNum varchar,
Low varchar,
Up varchar
);
Some statistics for the Patent table
SELECT gyear, COUNT(*) FROM patents GROUP BY gyear ORDER BY gyear; gyear | count -------+-------- 1975 | 72000 1976 | 70226 1977 | 65269 1978 | 66102 1979 | 48854 1980 | 61819 1981 | 65771 1982 | 57888 1983 | 56860 1984 | 67200 1985 | 71661 1986 | 70860 1987 | 82952 1988 | 77924 1989 | 95537 1990 | 90364 1991 | 96513 1992 | 97444 1993 | 98342 1994 | 101676 1995 | 101419 1996 | 109645 1997 | 111984 1998 | 147574 1999 | 153593 2000 | 157596 2001 | 166065 2002 | 167424 2003 | 169105 2004 | 164413 2005 | 143927 2006 | 173922 2007 | 157502 2008 | 157894 2009 | 167537 2010 | 219909 (36 rows)
SELECT appyear, COUNT(*) FROM patents GROUP BY appyear ORDER BY appyear;
appyear | count
---------+--------
1901 | 3
1902 | 1
1903 | 4
1904 | 4
1905 | 1
1908 | 1
1909 | 1
1915 | 1
1918 | 1
1921 | 1
1922 | 1
1925 | 1
1929 | 1
1930 | 2
1931 | 2
1933 | 1
1936 | 2
1938 | 4
1939 | 1
1940 | 3
1941 | 3
1942 | 1
1943 | 4
1944 | 19
1945 | 18
1946 | 11
1947 | 4
1948 | 8
1949 | 10
1950 | 9
1951 | 11
1952 | 17
1953 | 12
1954 | 30
1955 | 33
1956 | 33
1957 | 32
1958 | 39
1959 | 56
1960 | 73
1961 | 85
1962 | 111
1963 | 110
1964 | 147
1965 | 167
1966 | 193
1967 | 230
1968 | 337
1969 | 584
1970 | 1161
1971 | 2658
1972 | 9418
1973 | 41655
1974 | 65683
1975 | 65902
1976 | 65813
1977 | 65999
1978 | 65613
1979 | 65729
1980 | 66500
1981 | 63930
1982 | 65029
1983 | 61578
1984 | 67085
1985 | 71463
1986 | 75108
1987 | 81499
1988 | 90178
1989 | 96139
1990 | 99372
1991 | 100262
1992 | 103926
1993 | 108346
1994 | 123358
1995 | 144655
1996 | 144864
1997 | 169578
1998 | 168471
1999 | 180655
2000 | 196639
2001 | 208908
2002 | 206229
2003 | 191514
2004 | 181840
2005 | 166524
2006 | 139318
2007 | 97911
2008 | 48539
2009 | 12245
2010 | 1052
| 2
(91 rows)
patentdata=# SELECT COUNT(*) FROM patents; count --------- 3984771 (1 row)
patentdata=# SELECT kind, COUNT(*) FROM patents GROUP BY kind ORDER BY kind;
kind | count
------+---------
| 2302931
A1 | 1
B1 | 561885
B2 | 1119954
(4 rows)
SELECT p.patent, COUNT(c.patent) AS numcited FROM patents AS p, CITATIONS AS c where p.patent=c.citation GROUP BY p.patent ORDER BY numcited DESC; patent | numcited ---------+-------------- 4683202 | 1992 4723129 | 1935 4683195 | 1814 4463359 | 1676 4740796 | 1638 4558333 | 1558 4345262 | 1537 4313124 | 1504 4459600 | 1461 4733665 | 1286 5103459 | 1118 5572643 | 1018 4901307 | 959 5143854 | 924 5523520 | 918 5643826 | 883 4655771 | 861 4340563 | 808 5742905 | 792 5892900 | 780 4799156 | 771 4816567 | 751 5172338 | 749 4886062 | 745 4776337 | 744 4965188 | 742 4800882 | 726 4580568 | 706 5710887 | 697 4665906 | 697 3953566 | 681 5056109 | 678 4405829 | 666 4739762 | 665 4503569 | 662 4608577 | 659 4179337 | 642 5794207 | 636 5064435 | 635 5530852 | 630 5272236 | 629 4100324 | 625 5708780 | 623 5608786 | 620 5109390 | 618 5715314 | 616 5923962 | 610 5101501 | 608 5774660 | 607 4908112 | 607
