Welcome to Patentopia! Patentopia extracts patents from the USPTO database given one of two sets of inputs: 1) Principal Investigators (PIs), i.e., individuals; or 2) Firm names. A window in time can be selected in either forward or backward direction; if forward, this allows the patents to be an outcome after a given date; backward allows you to use the patents as a control, predictor, or instrumental variable prior to a given date. This is designed to be run on an observation level – for instance, a survey response date, company launch date, SBIR award date, or another reference. If no dates are listed in your dataset, Patentopia searches all available patents for all time.
Patentopia is designed to study US innovation and thus “all” patents refers only to US patents.
You will have the option to select:
- Inventor (Principal investigator) or Firm name (you must pick one)
- Forward or backward search – are you looking for a fixed time starting with a given date, or for a window prior? (you must pick one)
- Time window, defined in years and months, with respect to a given date (i.e., a survey response date, award or grant date, or another observation-level date of interest). Entering 0 years/0 months on the input screen causes Patentopia to look for patents for all time. Similarly, if you do not enter a reference date in the input file, Patentopia will look for patents for all time.
All files should be in the form of a comma separated value (CSV) file. Excel sheets should be saved as CSV files. You may use these templates: pi_input_template and firm_input_template . Alternatively, if you are working in excel with your own data, please see these instructions for saving your CSV file on a Mac or saving your CSV file on a Windows machine .
Please use the column headings as indicated for both PI or firm search.
- Principal investigators (individuals)
Patentopia was originally constructed to identify inventors associated with SBIR Principal Investigators (PIs). The name column should include all elements of the name (i.e., first, last; or first, middle, last) in a single cell. Patentopia will search for all patents in which the test name has a degree of overlap with the inventor name (USPTO field code: IN) and will generate a list of patents for names that are similar, with a fuzzy matching number specifying the degree of overlap between the name from the sample data and the target.
- Firm names
Patentopia can also be used to find companies who have been assigned patents by searching for all patents in which the assignee name (USPTO field code: AN) overlaps. It is not necessary to include incorporation flags such as LLC, Corp., Inc., etc.; but including them does not impact the code.
Note that if the company has patents through a separate agreement (for instance, through a license from a university), Patentopia will not find it because the university is still the assignee.
Many challenges exist in searching the USPTO database. Name matches are challenging as names may be misspelled, or many matches are found for common names. Geographical matches (i.e., by state) are challenging because people generally do not live at work! – for instance, someone may live in Washington, DC and work in Maryland.
Therefore, our process includes matching measures to estimate the overlap on two fronts: 1) between inventor names; 2) US state names. This is just a guide (but faster than using the USPTO search function on a case-by-case basis!).
Send comments and suggestions here:
The complete output code books are contained in Patentopia codebook 12-May-20. You will be emailed three zipped files:
- Sender “Patentopia team”
- Subject “Patentopia Data Results xxx” where xxx is a time stamp in Pacific Time.
All output files are labeled with username and the time staamp
- Complete output – lists all relevant patents individually and matching measures: Each row represents one patent for the firm or PI
- Abridged output – aggregates patents based on the matches: Each row represents all patents for that firm or PI.
- Timeout – (2 files) list PI or firm name where Patentopia timed out (i.e., the process was incomplete); and list patents at the moment of timeout. If this file is empty, then all observations were analyzed completely.
These files are labeled with your username and the time so you can identify the specific run.
The data fetch typically takes a few minutes for a file of ~100-500 records, but as this is hosted on AWS, the system may take even longer before sending you an email. Typically the firm search is faster than the PI search because of fewer disambiguation issues.
Please send us your feedback and experience with run times to LosAngelesNode at gmail dot com.
Send emails with your questions to LosAngelesNode at gmail dot com. You may also email if you do not receive an output within 24 hours. When communicating, please:
- Indicate the registered user ID on Patentopia
- Please forward any error messages that you receive as they have the time stamps.