This project is superceded by https://github.com/rendall/hometown-parser-2
Input a string, output an array of possible place-names. Demo: https://rendall.github.io/hometown-parser/
The 'hometown' field for Burning Man theme camps is often creatively interpreted by users filling out the form, so location data can be difficult to glean. This is an attempt to normalize the information. It may be useful in other similar circumstances as well.
This is a hand-crafted algorithm based on pattern-matching and trial-and-error. Still, even the most creative essays and interpretations of 'hometown' in this field follows general patterns, and the algorithm is good enough.
- incorporate the World Cities dataset
You are hosting a gathering of creative people from all over the world, and in order to be well prepared, you have an online form by which affiliated individuals let you know they are affiliated. Your form perhaps has a 'group name' field and a 'date of arrival' field and a 'number of members' field.
You also have a 'hometown' field.
For some of these groups, it's pretty straightforward:
"Moscow"
"Helsinki"
"Peoria, IL"
But for other groups, it's complicated: "Sydney, Salt Spring Island, BC and Christiania, Denmark" "SEATTLE/LOS GATOS/NYC" "Phoenix, Wyszków"
Other folks like to write novels: "Goa. This is were the founding members of the group met, most people are from different countries but Goa is the origin". "Vatican City (mostly), with a couple from Mauritania, Utah, and Zion, IL"
There are also many names for towns, and folks use them all: Montreal, Montréal San Francisco, San Francisco Bay Area, SF, SFO Philadelphia, Philly, PHL
You would love to, somehow, get a good grasp of where all these groups are coming from, but "SF" is so different from "RIO DE JANEIRO-BRAZIL AND SAN FRAN"
This is where the Hometown Parser can help you out.
The parser uses Ramdajs. You will need to reference that before loading hometown.js. Then reference hometowns.js.
If you're making a website, you can try putting this into the <head> tag:
<script src="https://cdnjs.cloudflare.com/ajax/libs/ramda/0.24.1/ramda.min.js" integrity="sha384-OricteDxGJ77wQTPXdPmz8SoLVmEkW51fQRzhTGmUXG6u608OIFWHzCVwDFki+w6" crossorigin="anonymous"></script>
followed by
<script src="https://raw.githubusercontent.com/rendall/hometown-parser/master/js/hometowns.js"></script>
but, I dunno. You might want to have those files be local, instead.
Alternatively, you can clone the repository which contains all that is necessary. Therein is a hometown.html webpage with examples and unit tests.
Run your string through the parseTowns function, as in
parseTowns("Our group members hail primarily from: San Francisco, Seattle, Louisville, Tolendo, Toronto, Vegas and Southern California")
and the Hometown Parser will return an array of strings representing its best guess as to what those place names are:
["San Francisco","Seattle","Louisville","Tolendo","Toronto","Vegas","Southern California"]
You can also then run those strings through toCanonicalName as in
parseTowns("NYC/PHL/PDX/OKC/LA/Vegas").map((s) => toCanonicalName(s))
which will output
["NYC","Philadelphia","Portland","Las Vegas","Oklahoma City","Los Angeles","Las Vegas"]
Check hometown.html for examples.
Definitely! Create an issue or:
Clone the repo and get to work! It's written using Typescript and Ramda, so if you know / want to learn those, this is a good project for you.
The algorithm is by no means perfect, so if you see a way to improve it, show me!
- Install NPM
- Clone this repository.
- Navigate to the repo directory using your favorite terminal.
- Type
npm install. ./ts/hometown.tsis the working copy. Edit that.- Type
npm build. - Open
hometown.htmlin your favorite browser to test.