Difference between revisions of "Redesign Assignment and Patent Database"

From edegan.com
Jump to navigation Jump to search
Line 58: Line 58:
 
All columns (variables): Two possibilities, depending on whether we want data to be repeated or not.
 
All columns (variables): Two possibilities, depending on whether we want data to be repeated or not.
 
 
Not-repeated (with exception of primary/foreign key): rf_id, next_transaction (also rf_id), num_conveyance (number in order of conveyance by date), acknowledgment_date (in ASSIGNOR for the assignment data paper but would be moved here. Could originally be stored in ASSIGNOR until CONVEYANCE is generated), execution_date (same deal as acknowledgment date, should probably be moved here; if there are multiple execution dates, we will store the last one) conveyance_type
+
Not-repeated (with exception of primary/foreign key):  
 +
* rf_id (bigint) reel frame number
 +
* next_transaction (bigint) rf_id of next transaction
 +
* num_conveyance (int) number in order of conveyance by date
 +
* acknowledgment_date (date) date that the assignor acknowledged the transaction
 +
* execution_date (date) date that the transaction
 +
* conveyance_type (varchar(255)) the USPTO paper determined a couple of different conveyance types and also outlined how to determine based on the conveyance text in the xml file which conveyance type is relevant to the transaction. The conveyance types are:
 +
** assignment
 +
** merger
 +
** change of name
 +
** government interest
 +
** agreement
 +
** security agreement
 +
** release
 +
 
 
 
Repeated, also include: assignee, assignor, patent_id
+
Repeated, also include:  
 +
* assignee_name (varchar(255)) name of the assignee
 +
* assignor_name (varchar(255)) name of the assignor
 +
* patent_number (varchar(255)) identifying number for the patent
  
 
===DOCUMENT_INFO===
 
===DOCUMENT_INFO===

Revision as of 16:50, 29 September 2017


McNair Project
Redesign Assignment and Patent Database
Project logo 02.png
Project Information
Project Title Redesign Assignment and Patent Database
Owner Shelby Bice
Start Date 9/2017
Deadline
Keywords patent
Primary Billing
Notes
Has project status
Is dependent on Reproducible Patent Data
Copyright © 2016 edegan.com. All Rights Reserved.


This is an extension of the work I did last semester under "Redesigning Patent Database". Instead of simply reconfiguring the existing database, this project encompasses and full redesign and creation of a new Patent database and a new Assignment database that will be joined together.

Assignment Database Structure

After reading through the paper "The USPTO Patent Assignment Dataset: Descriptions and Analysis"(https://www.uspto.gov/sites/default/files/documents/USPTO_Patents_Assignment_Dataset_WP.pdf) and taking into account are needs for the Assignment database (the ability to connect the two databases together, the ability to trace conveyance of a patent over time) I decided on the following structure for the Assignment database.

The foreign key that connects every table in the database is rf_id, which stands for reel frame id. It is unique for each entry in the main table, assignment and is included in every table.

Conveyance is the only table that cannot be fully constructed as data is inserted into the database. Some of the fields depend on other tables being constructed, so it may be either partially populated while the other tables are being populated or is the last table to be populated.

Full list of tables: assignee, assignor, document_info, assignment, conveyance

ASSIGNEE

Purpose: store all the assignees from all the assignments

Each entry represents: the recipient of a patent in an entry in the ASSIGNMENT table

Primary key: rf_id, unless it is possible to have multiple assignees, then rf_id and assignee name

All columns (variables) in the table:

  • rf_id (bigint) reel frame number
  • assignee_name (varchar(255)) name of the assignee
  • address_line_1 (varchar(255)) first line of address of assignee
  • address_line_2 (varchar(255)) second line of address of assignee
  • city (varchar(255)) city of assignee
  • state (varchar(255)) state of assignee
  • country var(char(255)) country of assignee
  • postal_code (varchar(255))

ASSIGNOR

Purpose: store all the assignors from all the assignments

Each entry represents: the one assigning/granting the license for the entry in the ASSIGNMENT table

Primary key: rf_id and assignor_name

All columns (variables):

  • rf_id (bigint) reel frame number
  • assignor_name (varchar(255)) name of the assignor

CONVEYANCE

Purpose: represent the conveyance of a particular patent overtime

Each entry represents: an individual transaction for a patent (not unlike assignment, except this table has information about the assignment in relation to other assignments for the patent

Primary key: rf_id

All columns (variables): Two possibilities, depending on whether we want data to be repeated or not.

Not-repeated (with exception of primary/foreign key):

  • rf_id (bigint) reel frame number
  • next_transaction (bigint) rf_id of next transaction
  • num_conveyance (int) number in order of conveyance by date
  • acknowledgment_date (date) date that the assignor acknowledged the transaction
  • execution_date (date) date that the transaction
  • conveyance_type (varchar(255)) the USPTO paper determined a couple of different conveyance types and also outlined how to determine based on the conveyance text in the xml file which conveyance type is relevant to the transaction. The conveyance types are:
    • assignment
    • merger
    • change of name
    • government interest
    • agreement
    • security agreement
    • release


Repeated, also include:

  • assignee_name (varchar(255)) name of the assignee
  • assignor_name (varchar(255)) name of the assignor
  • patent_number (varchar(255)) identifying number for the patent

DOCUMENT_INFO

Purpose: store extra information relevant to the patents represented in ASSIGNMENT

Each entry represents: information about the documentation and the patent for a particular entry in ASSIGNMENT

Primary key: rf_id

All columns (variables) when created: rf_id, app_num (application document USPTO number), app_date, app_country, pgpub_num (pre-grant publication document USPTO number), pgpub_date, pgpub_country, grant_num (granted patent document USPTO number, also know as the patent number), grant_date, grant_country, invention_title, language (of invention title – could be potentially useful and interesting to investigate), reel_num, frame_num (reel_num and frame_num were in ASSIGNMENT for the uspto data paper, but I think they would be more useful here)

ASSIGNMENT

Purpose: represent assignment transactions – central table for the database

Each entry represents: an individual patent assignment transaction

Primary key: rf_id

All columns (variables): rf_id, assignment_id (US patent assignment id), correspondent_name, correspondent_address1, correspondent_address_2, correspondent_address3, correspondent_address4, record_date (date recorded with USPTO), last_update_date, page_cnt (page count of assignment record)

More extensive notes exist under :E/McNair/Projects/Redesigning Patent Database/New Patent Database Project/Notes on USPTO Assignment Data Paper

Patent Database Structure

Most of this design is based off what I investigated last semester and therefore is subject to change, since the design for the Assignment database is somewhat new.

The foreign key that connects every table is patent_number, which uniquely identifies every patent in the PATENT table and is included in every table.

PATENT

Purpose: represent patents – central table for the database

Each entry represents: an individual patent

Primary key: patent_number

All columns (variables): patent_number, grantdate, prioritydate, prioritycountry, prioritypatentnumber, cpcsubgroup, pctpatentnumber, claims, appnum, grantyear, appdate, appyear, nber, uspc, uspc_sub

ASSIGNEE

Purpose: Represent the person the patent is currently assigned to according to the patent information (the information in the Assignment database is more complete)

Each entry represents: the assignee for a patent in PATENT

Primary key: patent_number, first_name, last_name (in case it is possible for there to be multiple assignees for an individual patent)

All columns (variables): First name, last name, address, postcode, orgname, city, country, patentnumber, state, residence, size of firm

CITATION

Purpose: Represent all the patents in PATENT that another patent in PATENT cites

Each entry represents: one citation that a patent in PATENT makes

Primary key: patent_number AND cited_patent_number (because a patent might cite multiple patents, so both variables are needed for the primary key)

All columns (variables): patent_number, cited_patent_number

FEE

Purpose: Represent information about the fees paid for an individual patent

Each entry represents: information about the fees paid on the patent

Primary key: patent_number

All columns: patent_number, four, eight, twelve, fee_date, fee_code

HISTPATENT

Purpose: Represent historical data about the patent pertaining to dates and publication

Each entry represents: sequence of historical data about an individual patent in PATENT

Connecting Patent database and Assignment database

The answer to connecting the Patent database to the Assignment database lies somewhere in using the information in the table DOCUMENT_INFO to connect to a patent_id from the Patent database PATENT table for each assignment in the Assignment database ASSIGNMENT table. On further investigation into "The USPTO Patent Assignment Dataset: Descriptions and Analysis, the field they called "grant_doc_num" (grant_num above) in DOCUMENT_INFO is the patent number (the description of DOCUMENT_INFO above has been altered to reflect this.)

The paper mentions that there will errors in patent number stored in DOCUMENT_INFO under grant_doc_num so a separate table called DOCUMENT_INFO_ADMIN was constructed to determine how prevalent errors were. They queried the patent number from for the appno_doc_num from administrative data. The "appno_doc_num" is the Application Document USPTO number from DOCUMENT_INFO. On further investigation into the paper, it became clear that "administrative data" refers to internal USPTO data that was available to the authors of the paper, but that we might not have. However, in 99% of cases the grant_doc_num and the queried patent number based on the appno_doc_num were the same, so we can probably rely on grant_doc.

Therefore, since every table in the Patent database will have the patent number stored, to connect any table in Assignment to the Patent database, it would first be joined with DOCUMENT_INFO on rf_id and then joined with the appropriate table in Patent on patent number.

Also in the paper, they mentioned the Assignments on the Web for Patents (AOTW-P), a searchable database of individual USPTO assignment records keyed on reel-frame identification, patent number, and assignor or assignee name (https://assignment.uspto.gov/patent/index.html#/patent/search). Obviously it would not be possible to individually use this tool to query all the patent numbers, but if it would be possible to write a script to somehow query each patent number using the rf_id and parse the response, this could potentially be useful to check the patent numbers, but might not be any more accurate than what will already be in DOCUMENT_INFO.