It’s faster and easier to add the location of your local Starbucks, Sparkasse, Сбербанк, or ファミリーマート to OpenStreetMap: just add a point and start typing a name, and the editor will autocomplete the rest. We added support for thousands of common restaurants, supermarkets, banks, clothing stores, and other chains.
This also improves the quality of the data people are adding since no one needs to know that by convention OpenStreetMap classifies Domino’s Pizza as a fast food joint rather than a restaurant, or whether or not it should have an apostrophe — iD’s preset system takes care of that for you.
This new preset database is itself built using OpenStreetMap data: we mine OpenStreetMap for the most common POIs and extract the correct tagging and spelling of their names. Thus it’s a two-way street — the presets will both help improve the quality of OpenStreetMap, and become more complete and accurate as OpenStreetMap improves.
Read on for the technical details.
The traditional way of mapping a feature on OpenStreetMap is to create the geometry (node, way, or relation) and then tag it with values like amenity=fast_food and name=Carl’s Jr. This requires the user to know the OpenStreetMap tags that exist and pick the best one for the type of feature they want to map. That can be a hindrance for new users who don’t know the tags or in certain situations where something could be tagged in several different ways.
We made the name-suggestion-index with the goal of creating a canonical dataset that can be used for suggesting correct spelling and tagging in OpenStreetMap. The dataset is generated by going through the OSM planet, grouping names with their incorrect counterparts, and keeping track of how often they are used. Determining which names are correct and incorrect just comes down to usage: whatever is used most is considered the most correct. After processing, we have an index of about 1,400 of the most common values for the type of features we’re interested in.
iD uses this index in two ways. First, when a user edits the name of a feature, iD autocompletes common names. For example, if a feature has been tagged as fast food and a user begins typing “McD…”, iD suggests McDonald’s. Second, we generate presets from the index. Presets allow users to quickly create features without needing to know the underlying tagging scheme. This allows for quicker entry for the most common features, encourages correct tagging and naming, and helps teach new users by examples from a vetted list of values.