Difference between revisions of "Patent Data Specifications"
Line 5: | Line 5: | ||
The patent database contains the merged datasets from the USPTO bulk data and Harvard Dataverse using SQL. Specifics on how the datasets were merged are given in [[Patent Data Processing - SQL Steps]]. The patent database consists of 16 tables covering patents, patent litigation, patent maintenance, patent assignment, and other details on patent owners. Additionally, known issues with each table have been recorded. | The patent database contains the merged datasets from the USPTO bulk data and Harvard Dataverse using SQL. Specifics on how the datasets were merged are given in [[Patent Data Processing - SQL Steps]]. The patent database consists of 16 tables covering patents, patent litigation, patent maintenance, patent assignment, and other details on patent owners. Additionally, known issues with each table have been recorded. | ||
<section end="pds" /> | <section end="pds" /> | ||
− | |||
==Tables== | ==Tables== | ||
Revision as of 12:29, 13 July 2016
Return to Patent Data Wiki Page.
Contents
Background
The patent database contains the merged datasets from the USPTO bulk data and Harvard Dataverse using SQL. Specifics on how the datasets were merged are given in Patent Data Processing - SQL Steps. The patent database consists of 16 tables covering patents, patent litigation, patent maintenance, patent assignment, and other details on patent owners. Additionally, known issues with each table have been recorded.
Tables
Data Issues=====Citations Table=
The table has two columns, 'citingpatentnumber' and 'citedpatentnumber'. There are rows with 'citedpatentnumber' greater than 10000000. For instance:
citingpatentnumber | citedpatentnumber --------------------+------------------- 9226901 | 102005013726 9226905 | 101332187 9226905 | 2006528175 9226905 | 2011513304 9226905 | 1020090061010 9226905 | 1020110049808 9226905 | 2010126349 9226909 | 101340916 9226909 | 10128910 9226914 | 102318827 9226914 | 2014109862 9226915 | 10111049 9226918 | 2008005345 9226918 | 2008005345 9226918 | 2008077092 9226918 | 2008077092 9226918 | 2008070268 9226918 | 2008128126 9226921 | 2008129994 9226922 | 2012000595 9226922 | 2012058127 9226923 | 2010135524 9226930 | 102002040 9226930 | 19507522 9226930 | 50106981 9226930 | 53082783 9226930 | 57052334 9226930 | 10029979 9226930 | 10045750 9226931 | 2008203212 9226931 | 2010227111 9226931 | 2012068515 9226931 | 2013106565 9226932 | 103458880 9226933 | 2006257751 9226933 | 101366734 9226933 | 101396048 9226933 | 101926831 9226933 | 20202562 9226933 | 202005009120 9226933 | 61063618 9226937 | 11510473 9226937 | 2001506579 9226937 | 2001519791 9226937 | 2003119127 9226937 | 2003517831 9226937 | 2005099761 9226937 | 2006076681 9226937 | 2006078941 9226937 | 2006079021 9226937 | 2006094209 9226937 | 2006094210 9226937 | 2006094233 9226937 | 2006094235
allpatent_clone=# SELECT COUNT(*) FROM citations WHERE citedpatentnumber>10000000; count --------- 1411140 (1 row)
allpatent_clone=# SELECT COUNT(*) FROM citations WHERE citedpatentnumber IS NULL; count ---------- 23516667 (1 row)
allpatent_clone=# SELECT COUNT(*) FROM citations; count ---------- 97680838 (1 row)
Possible Solution
Jul 8, 2016: The blank citedpatentnumbers were created due to inconsistency between the original type and the type in the citation table (string to integer). The blank entries mostly correspond to publication number, non-U.S. patent number, and non-standardized patent number. The next step would be to recreate the table accounting for these issues. The U.S. publication number could be matched to the publications numbers in the histpatent table and be replaced by the corresponding patent numbers.
The foreign cited patents will be moved to a separate table.
Assignees Table
'Country' is missing.
allpatent_clone=# SELECT COUNT(*) FROM assignees WHERE country=' '; count --------- 2361543 (1 row)
allpatent_clone=# SELECT COUNT(*) FROM assignees WHERE country='unknown'; count ------- 3918 (1 row)
allpatent_clone=# SELECT COUNT(*) FROM assignees WHERE country IN ('unknown',' ') AND state=' '; count --------- 1851353 (1 row)
UPDATE: The source of the problem seems to be the Harvard Dataverse.
Assignees & assigneesUSU Tables
No information provided about the assignee. No entries for orgname or first and last names.
allpatent_clone=# SELECT COUNT(patentnumber) FROM assignees WHERE (orgname= OR orgname IS NULL) AND (firstname= OR firstname IS NULL) AND (lastname= OR lastname IS NULL); count -------- 344794 (1 row)
AssigneesUSU table was made with the following code:
SELECT orgname, patentnumber, country, firstname, lastname, state INTO assigneesUSU -- assignees in US,unknown, and blank entries FROM assignees WHERE country IN ('US', , 'unknown') OR (state IS NOT NULL AND state!=) ORDER BY orgname;
allpatent_clone=# SELECT COUNT(patentnumber) FROM assigneesUSU WHERE (orgname= OR orgname IS NULL) AND (firstname= OR firstname IS NULL) AND (lastname= OR lastname IS NULL); count -------- 344793 (1 row)