Bulk Patent Assignee Processing

From edegan.com
Jump to navigation Jump to search

USPTO Assignees Data

We would like to download and absorb data from this location on the USPTO website into our tables. The objective is to determine whether this dataset is better than the current version of our patent data (a combination of the data in the patent_2015 and patentdata databases.

Steps Followed to Extract the 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:

<patent-assignment>
       +<assignment-record>
       +<patent-assignors>
       +<patent-assignees>
       +<patent-properties>
</patent-assignment>

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:

  -<assignment-record>
      <reel-no>27132</reel-no>
      <frame-no>841</frame-no>
     -<last-update-date>
         <date>20160122</date>
      </last-update-date>
      <purge-indicator>N</purge-indicator>
         -<recorded-date>
             <date>20111027</date>
          </recorded-date>
        <page-count>2</page-count>
     -<correspondent>
          <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>
       </correspondent>
       <conveyance-text>ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS).</conveyance-text>
 </assignment-record>


Assignors

Here are the columns in the assignors table:

  • reel_no
  • frame_no
  • assignor_name
  • execution_date

The corresponding XML node is :

-<patent-assignors>
   -<patent-assignor>
      <name>WALKER, MATTHEW J.</name>
     -<execution-date>
         <date>20090512</date>
      </execution-date>
    </patent-assignor>
   -<patent-assignor>
      <name>OLSZEWSKI, MARK E.</name>
     -<execution-date>
         <date>20090512</date>
      </execution-date>
    </patent-assignor>
  </patent-assignors>

Assignees

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:

 -<patent-assignees>
   -<patent-assignee>
       <name>KONINKLIJKE PHILIPS ELECTRONICS N V</name>
       <address-1>GROENEWOUDSEWEG 1</address-1>
       <city>EINDHOVEN</city>
       <country-name>NETHERLANDS</country-name>
       <postcode>5621 BA</postcode>
     </patent-assignee>
   </patent-assignees>

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:

 -<patent-properties>
   -<patent-property>
     -<document-id>
         <country>US</country>
         <doc-number>14143589</doc-number>
         <kind>X0</kind>
         <date>20131230</date>
      </document-id>
     -<document-id>
         <country>US</country>
         <doc-number>20140260305</doc-number>
         <kind>A1</kind>
         <date>20140918</date>
      </document-id>
     <invention-title lang="en">LEAN AZIMUTHAL FLAME COMBUSTOR</invention-title>
   </patent-property>
 </patent-properties>

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">
    <action-key-code>DA</action-key-code>
   -<transaction-date>
       <date>20160122</date>
    </transaction-date>
   -<patent-assignments>
       +<patent-assignment>
       +<patent-assignment>
       +<patent-assignment>
       +<patent-assignment>
       +<patent-assignment>
       +<patent-assignment>
            .
            .
            .
     </patent-assignments>
 </us-patent-assignments>


DTD

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 city (#PCDATA)>
<!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 kind (#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)*>
]>


Scripts for processing data

The programs/scripts are located at our Software Repository.

repository: Patent_Data_Parser 
branch: next
The scripts are in uspto_assignees_xml_parser directory.


Downloading raw bulk data from USPTO

The XML files are available at https://bulkdata.uspto.gov/data2/patent/assignment/

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 (see above).


Parsing the XML files

Inserting Extracted Data into Tables

Clean Up