Say someone asks for help getting a list of utility patents from Los Angeles, what would you do? You can’t use the USPTO’s peds api, it does not offer a search by location. You could use the USPTO’s patft, a search for ic/”Los Angeles” and is/CA and apt/1 would do the trick (ic/ matches the inventor’s city and is/ matches the inventor’s state and apt/1 matches the application type of utility patents). The problem is that it only returns 50 patents at a time (29,148patents met those conditions at the time this was written). There isn’t a download option so your mouse finger would get quite tired from all the clicking you’d have to do.
One solution would be to use the query tool provided by the patentsview api! Like the USPTO’s patft and peds searches, it will return data on patents issued in 1976 onward. It takes a couple of screens to enter what you want, but if you stick with it, it will email you a link to a csv file (or json file) of the patents that met your criteria.
Here’s what I just did: Click on the Advanced Search link on the query tool page. Under the Patents section set Patent Type equals Utility and then click on the +Add to Search link. Under the Inventors section set the fields to “Inventor Location At Issue equals United States (country) California (state) Los Angeles (City)” and then click the +Add to Search link. Then click the Submit Search link.
On the subsequent screen you get to pick the fields you want returned. I selected my favorite fields as the request for help didn’t specify any.
Click on the Preview Query when you’ve selected all the fields you want. You then can specify a sort order, enter your email address, prove you are not a robot and click on the Submit Query link. That’s it! A short while later an emailed arrived with a link to my csv file.
A couple of things to point out about my field choices: I included the inventor sequence. There can be, and usually are, multiple inventors on a patent from potentially different cities or even different countries. Each inventor will be on a separate row in the csv file, so a patent with five inventors will have five rows in the csv file. A sequence of 0 indicates the first inventor on the patent. I could have added that to my search criteria as shown below. Again, what constitutes a patent from Los Angeles was not specified (any inventor from there or only if the first inventor hails from there or possibly an assignee from there). I also selected the pre disambiguated names as sometime the api would massage the names in an effort to be helpful. (They try to figure out if John Doe on one patent is the same person as John Q Doe on another patent; if they think they are, they’d change the data and use one name consistently.)
Another thing to point out is that this only includes inventors where the USPTO’s city field is Los Angeles (the patentsview database is built from bulk patent files the USPTO makes available to anyone). There are patents with a city of “late of Los Angeles” (to indicate a deceased inventor) or Los Angeles County that would not be included in the query I made. I’d have to do separate queries and merge the csv files to include these patents.
One important api limitation to point out is that it will only return 100,000 patents, the result set will be silently truncated. We’re fine in this case, we were safely under the limit. If you search for Toyko, Japan however you would reach the limit. You’d have to make multiple queries using the patent’s grant date for example (add in that the issue date was before a specific date and then a separate query adding in that the issue date was greater than or equal to that date). You’d have to play around with each query to make sure that less than 100,000 patents are returned.
Also of note, the patentsview database is updated roughly quarterly while the USPTO’s patft is updated weekly. Your patentsview results may not contain the most recently issued patents. Oh, and as one last, slightly troubling caveat, the patentsview database contains all the patents in the bulk files just mentioned, even around 8,000 patents which were withdrawn after issue. This means that there would be a small chance (8,000 withdrawn patents are included in the database of roughly five million patents) that some of the patents in your csv file have been withdrawn. The USPTO’s patft does not return data for patents that have been withdrawn. I’ve pointed this out to the patentsview team but they haven’t taken any action yet.