Roads about …

Why

On a road trip across the country I noticed "501st Ave" intersecting our little highway a couple hours West of Minneapolis. The numbers had been steadily increasing as we drove further West. I thought "That doesn't happen in North Carolina!" It got me thinking about how patterns of roads differ around the country. When we got home, I created maps of alphanumeric gridded roads showing these alphanumeric patterns for several States. Its different, its interesting, someday maybe I'll make a national map.

I've also noticed roads that fit a theme: "Death Star Lane" in my hometown (sci fi fan!). "Convertable Lane" on another road trip (gear head!). "Canis Minor Road" the next town over (astronomy buff!).

Wouldn't it be cool to be able to find all the roads that Dead Head fans had memorialized around the country?

This is a way to do just that.

How it is made

A basemap is compiled offline: OpenStreetMap road geometry, attributed against US Census TIGER/Line county and state boundaries so every road carries a stable FIPS identifier. Names are normalized into a dictionary with hash-derived ids that stay put across TIGER vintages, then cut into vector tiles.

Datasets of road names and FIDS ids are generated, along with vector embeddings or the road names.

Themes are discovered by reviewing the most popular articles on wikipedia via QRank data. From there, I use different queries against the Wikidata Query Service to find articles of interest.

What it cannot tell you

  • Just because a name matches our theme, it doesn't mean that the author intended it: an 'Elvis Drive' might just be named after someone's uncle.
  • OpenStreetMap street coverage is uneven; a quiet county may be under-mapped rather than under-named. (But its a lot better than TIGER data!)

Credits

Data sources:

Tools:

  • Haiku: theme candidate disambiguation
  • Astro: site framework
  • SST: infrastructure deployment
  • AWS: hosting infrastructure

Email me if you have questions, comments, or suggestions.