Global stores taxonomy

Introduction

The stores taxonomy contains a list of locations where products (food products, beauty products, etc.) can be obtained.

Store entries

An entry in the taxonomy describes a single location. This taxonomy can combine data pulled from various sources: receipts from prices, osm, wikidata, etc or provided by the user (honour system). OFF should be careful not copy to much data already maintained by other parties. And if data is copied, it is to provide a snapshot in time.

Receipts

Example of store information on a receipt

The basis could be the receipts that users have added on prices. As all OFF data, any information should be backed up by image proofs. In the case for a store, this could be a receipt. The store information available on the receipt can be extracted to the taxonomy. What can be extracted depends on the receipt. In the example shown here, it is possible to extract the store name, the operator (useful for franchises), the address, phone number, siret, naf and tva. Analysus of more receipts might reveal other country specific information. Hopefully this information is enough to unambiguously identify a store.

Prices

The possible fields used follow the data structure used by OFF Prices and describes a single location:

id - a unique identifier (as defined by OFF Prices)
osm_id - the corresponding node(only?) in OpenStreetMap
name - a long name
postcode - the postcode
city - the city
country - the country tag
lat - the latitude
lon - the longitude

Example

An example for an entry / location / store / shop in the taxonomy: osm_id:en: [[1]]
osm_type:en: NODE
price_id:en: 45
osm_name:en: Biocoop
osm_display_name:en: "Biocoop, Rue de Tolbiac, Quartier de la Maison-Blanche, Paris 13e Arrondissement, Paris, Île-de-France, France métropolitaine, 75013, France"
osm_address_postcode:en: 75013
osm_address_city:en: en:Paris
osm_address_country:en: en:france
osm_lat:en: 48.8259597
osm_lon:en: 2.3511541

Open Street Map integration

All locations are linked to Open Street Map (OSM) through the unique OSM id. The link refers to a basic object in SOM:

Node - is used to refer to a single location, like BP Service;
Way - is used to refer to a building, like IKEA;
Relation - is used to refer multiple ways used by the shop, like Carrefour;

Using a Way to designate a shopping location might lead to inconsistencies, like this actual location. Which shop on the street is meant? Using a relation might lead to inconsistencies as well, like this actual location. This can be limited to OSM objects that have the tag shop.

Several of the fields are pulled from OSM (name, postcode, city, country, lat, lon). These need to be regularly updated, so that OSM is followed. Any deleted location should be kept.

Maintenance

The stores taxonomy needs to be regularly updated from the data entered through OFF Prices. This means any changed values and any added entries/

Discussion points

What information should be stored by OFF and what can be left to OSM or Wikidata?
What to be done with shops that go out of business?

Note

We need to associate stores with brands that are sold in thoses stores (Auchan for Auchan…)
We should try to add Wikidata relations (and thus OSM ones) whenever possible
Chains - if a store is part of a chain with multiple stores it can be added through a wikidata link

Moved to GitHub: https://github.com/openfoodfacts/openfoodfacts-server/blob/master/taxonomies/stores.txt