Bulk Patent Assignee Processing

From edegan.com
Jump to navigation Jump to search

Bulk Patent Assignee Processing
Project logo 02.png
Project Information
Has title USPTO Bulk Data Processing
Has start date
Has deadline date
Has keywords Data
Has project status Subsume
Has sponsor McNair Center
Has project output Data, Tool, How-to
Copyright © 2019 edegan.com. All Rights Reserved.

Return to Patent Data.

The USPTO provides bulk data recording patent transactions, applications, properties, reassignments, and history through XML files to the general public. These files have been downloaded and the data has been compiled in tables using PostgreSQL. The objective of processing the bulk data is to enhance the McNair Center's historical datasets (patent_2015 and patentdata) and track the entirety of US patent activity, specifically concerning utility patents.

Steps Followed to Extract the USPTO Assignees Data

Extracting Data from XML Files

All the historical USPTO data is available as XML files. Here is the tree structure for the XML files:


Each of the above internal nodes is mandatory, and is a logical grouping of information fields. Each node has a corresponding table created with more or less the same fields as the XML elements.

Corresponding tables are:

  • assignment-records : assignment
  • patent-assignors : assignors
  • patent-assignees : assignees
  • patent-properties : properties

Additionally, for each file that is downloaded, there are some associated specs. All of these are stored in the PatentAssignment table. Here is the data model diagram.

Assignment Records

The fields in the assignment record are:

  • last_update_date
  • purge_indicator
  • recorded_date
  • correspondent_name
  • correspondent_address_1
  • correspondent_address_2
  • correspondent_address_3
  • correspondent_address_4
  • conveyance_text

Here is the corresponding XML that we are mapping:

          <name>DOUGLAS B. MCKNIGHT</name>
          <address-1>595 MINER ROAD</address-1>
          <address-2>INTELLECTUAL PROPERTY & STANDARDS</address-2>
          <address-3>CLEVELAND, OH 44143</address-3>
       <conveyance-text>ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS).</conveyance-text>


Here are the columns in the assignors table:

  • reel_no
  • frame_no
  • assignor_name
  • execution_date

The corresponding XML node is :

      <name>WALKER, MATTHEW J.</name>
      <name>OLSZEWSKI, MARK E.</name>


Here are the columns in the assignees table:

  • reel_no
  • frame_no
  • assignee_name
  • assignee_address_1
  • assignee_address_2
  • assignee_city
  • assignee_state
  • assignee_country
  • assignee_postcode

The corresponding XML nodes are:

       <address-1>GROENEWOUDSEWEG 1</address-1>
       <postcode>5621 BA</postcode>

Patent Properties

Here are the columns in the properties table:

  • reel_no
  • frame_no
  • documentid
  • country
  • kind
  • filingdate
  • invention_title

The corresponding XML segment would be:

     <invention-title lang="en">LEAN AZIMUTHAL FLAME COMBUSTOR</invention-title>

Patent properties have a many-to-one relationship : one patent can have more than one properties.

Note: We are not sure what documents with kind 'X0' say

Patent Assignment

Every XML file download has some fields associated with it, in addition to a number of patent assignment nodes.

Here are the columns in the table:

  • reel_no
  • frame_no
  • action_key_code
  • USPTO_Transaction_Date
  • USPTO_Date_Produced
  • version

Here is what the XML in a downloaded file looks like:

 <?xml version="1.0" encoding="UTF-8"?>
 <!DOCTYPE us-patent-assignments>
-<us-patent-assignments date-produced="20131101" dtd-version="1.0">


Here is the DTD specified by the USPTO, which specifies optional fields and :

<?xml version="1.0" encoding="utf-8"?> 
<!DOCTYPE us-patent-assignments [<!ELEMENT us-patent-assignments (action-key-code, transaction-date, patent-assignments)>
<!ATTLIST us-patent-assignments  dtd-version   CDATA  #IMPLIED 
				 date-produced CDATA  #IMPLIED> 
<!ELEMENT action-key-code (#PCDATA)> 
<!ELEMENT transaction-date (date)>
<!ELEMENT patent-assignments (data-available-code | patent-assignment+)>
<!ELEMENT date (#PCDATA)> 
<!ELEMENT data-available-code (#PCDATA)> 
<!ELEMENT patent-assignment (assignment-record, patent-assignors, patent-assignees, patent-properties)> 
<!ELEMENT assignment-record (reel-no, frame-no, last-update-date, purge-indicator, recorded-date, page-count?, correspondent, conveyance-text)> 
<!ELEMENT patent-assignors (patent-assignor+)> 
<!ELEMENT patent-assignees (patent-assignee+)> 
<!ELEMENT patent-properties (patent-property+)> 
<!ELEMENT reel-no (#PCDATA)> 
<!ELEMENT frame-no (#PCDATA)> 
<!ELEMENT last-update-date (date)> 
<!ELEMENT purge-indicator (#PCDATA)>  
<!ELEMENT recorded-date (date)> 
<!ELEMENT page-count (#PCDATA)> 
<!ELEMENT correspondent (name, address-1?, address-2?, address-3?, address-4?)> 
<!ELEMENT conveyance-text (#PCDATA)> 
<!ELEMENT patent-assignor (name, execution-date?, date-acknowledged?)> 
<!ELEMENT patent-assignee (name, address-1?, address-2?, city?, state?, country-name?, postcode?)> 
<!ELEMENT patent-property (document-id*, invention-title?)> 
<!ELEMENT name (#PCDATA)> 
<!ATTLIST name name-type (natural | legal)  #IMPLIED> 
<!ELEMENT address-1 (#PCDATA)> 
<!ELEMENT address-2 (#PCDATA)> 
<!ELEMENT address-3 (#PCDATA)> 
<!ELEMENT address-4 (#PCDATA)> 
<!ELEMENT execution-date (date)>
<!ELEMENT date-acknowledged (date)> 
<!ELEMENT state (#PCDATA)>
<!ELEMENT country-name (#PCDATA)>
<!ELEMENT postcode (#PCDATA)> 
<!ELEMENT document-id (country, doc-number, kind?, name?, date?)> 
<!ELEMENT invention-title (#PCDATA | b | i | u | sup | sub)*>
<!ATTLIST invention-title  id   ID     #IMPLIED 
			   lang CDATA  #REQUIRED> 
<!ELEMENT country (#PCDATA)>
<!ELEMENT doc-number (#PCDATA)>
<!ELEMENT b (#PCDATA | i | u | smallcaps)*>
<!ELEMENT i (#PCDATA | b | u | smallcaps)*>
<!ELEMENT u (#PCDATA | b | i | smallcaps)*>
<!ATTLIST u  style  (single | double | dash | dots )  'single' >
<!ELEMENT sup (#PCDATA | b | u | i)*>
<!ELEMENT sub (#PCDATA | b | u | i)*>
<!ELEMENT smallcaps (#PCDATA | b | u | i)*>

Inserting Extracted Data into Tables

Clean Up

Scripts for processing data

The programs/scripts (see details below) are located on our Bonobo Git Server.

repository: Patent_Data_Parser 
branch: next
directory: /uspto_assignees_xml_parser

Downloading raw bulk data from USPTO

repository: Patent_Data_Parser 
branch: next
directory: /uspto_assignees_xml_parser
file: USPTO_Assignee_Download.pl

The down-loader script used to download XML files is essentially same, with minor changes, as the one used for downloading USPTO patent-data. That is, the current version of down-loader script downloads all files from the base URL: https://bulkdata.uspto.gov/data2/patent/assignment/

Parsing the XML files

repository: Patent_Data_Parser 
branch: next
directory: /uspto_assignees_xml_parser
file: uspto_assignees_XML_parser.plx


uspto_assignees_XML_parser.plx - Parses XML files and populates a database.

Specifically, parses every file in a directory according to a schema (see above). Then populates a database on the RDP.


perl uspto_assignees_XML_parser.plx /path/to/directory_containing_XML_files


Arguments The full path to directory is provided as a command line argument. It should contain the XML files to parse and no other file. This path should be specified in Windows format (with '\') and NOT unix format.

Features and Effects As each XML file is parsed, a database on local host (RDP) is populated. If at any point there is an error, for example a particular XML file is bad/invalid or the psql statement cannot be executed, the program aborts with a message.

We choose to populate local database because remote connections are too slow. The database is eventually moved to DataBase server manually.


The first version does the job as expected. It was used to populate the assignees database by parsing XML files from USPTO(see above). We parsed all XML files dated till 7/4/2016.


  • Add more command line options to improve usability.
  • Improve portability to allow Unix/Linux pathnames. This is straightforward to do with Perl modules File::Basename and File::Spec.