search

foursquare / fsqio

A monorepo that holds all of Foursquare's opensource projects
Apache License 2.0
254 stars 54 forks source link

Content Issues #39

Open kuroshhashemi opened 7 years ago

kuroshhashemi commented 7 years ago

First thanks to the Foursquare team for opening such a great resource. I've uncovered content issues, realizing some are TwoFishes-relevant while others Geonames-relevant

Countries

  1. East Germany should not be a country. Even shows on Foursquare.com... strange
  2. Democratic Republic of Congo and Republic of Congo are being treated as the same country in foursquare.com and latest index build. Strangely, the TwoFishes demo recognizes them as two different countries as they should be
  3. Cities in American Samoa follow wrong nomenclature. For example Pago Pago is labeled as Pago Pago, Eastern District instead of Pago Pago, American Samoa. (eastern section is a division within American Samoa)
  4. To be consistent all or no territories should be included. Currently only some territories are included. Those missing include American Samoa (AS), Antarctica (AQ), Bouvet Island (BV), French Southern Territories (TF), Isle of Man (IM), Tokelau (TK). I omit Western Sahara from this list because it is disputed however the same can be said of Antarctica

Rankings

  1. Madrid, Colombia ranks above Madrid, Spain (even with local bias this seems incorrect)

Consistent Treatment of Woetype

  1. It seems impossible to truly isolate towns/cities because while most towns/cities are in woetype 7, some are also in woe type 10 (for example Westport, MA). But if we open applications to woetype=7,10. Then we run into a lot of duplication issues such as Johannesburg below

Same city, multiple ID’s (these are just a few examples. Sometimes within same woetype. Other times span multiple woetypes )

  1. Two instances of Johannesburg, South Africa (Johannesburg and City of Johannesburg). Nothing online indicates there is a parental administrative area over Johannesburg with the same name)
  2. Westport Township vs Town of Westport vs Westport, SD
  3. Town of Howell, NJ vs Howell, NJ

Bounding Boxes

  1. Bounding box for US and Russia seem to be entire globe on twofishes demo and latest index. Though foursquare site seems to correct these two bounding boxes