Some Use Cases for ATGeo

Hi!

I’m looking https://atgeo.org while working on my new site where I’m collecting and sharing out data around standup comedy. While putting this together, there have been a few things that I’d like to use ATGeo for that don’t yet seem available.

  1. Search assumes you search per each collection, where collections are one of OSM, 4SQ, or Overture. I’d like to be able to search for a place regardless of collection, then maybe see a scored or ordered list of results. e.g. Searching “the comedy store” or “the comedy store West Hollywood, CA” could give me the record that associates to the 4SQ, OSM, & Overture ID
  2. As an app that plans to have venue’s “claim” their location, I’d like to support some ATProto native way to endorse the correctness of location data. Or to correct it.

From a high level, it seems like the approach of ATGeo is to mirror existing data sources and give them records. For my purposes at least, this doesn’t seem super useful? (I could be missing something here)

What would be super useful and interesting from my POV is if we joined the 3 data sources somehow, and created our an ATProto derivation of the combined set that could position itself as a successor. This is what I think would exist from putting together #1 and #2 along with some continuous reading in from the other 3 sources + merging logic.

3 Likes

Thanks for posting your needs @tyler.fun

We don’t think posting 2TB+ of data to ATProto is correct - the source of truth is off network.

There could be an endpoint that queried all of the sources and returned one list to you - you / the user will still need to pick.

The IDs then are used to aggregate across all the records. I’ll have the venue in my PDS for events, beacon bits check ins, and anything else that uses the lexicon.

You or another AppView can perform aggregation using the unique IDs.

Claiming is very much going to be up to your business logic.

For example @trezy.codes has imported all games from IGDB and will then verify devs / publishers who I guess will then have the canonical copy in their account.

I think you’ll have a similar flow for venues / comedy clubs.

I don’t know if pre-aggregating “comedy club venues in the US” is a small enough dataset for you to aggregate up front. Feels like it might be?

@schuyler.info any thoughts on this?

3 Likes