Randy Meech


A New OpenStreetMap Geocoder With Elasticsearch


Users of mobile mapping applications expect a forward geocoder with autocomplete and flexible search results, as well as an accurate reverse geocoder. Recent versions of Elasticsearch offer both excellent geo support and a fast completion suggester.

Pelias is a new open-source project with tools for indexing OpenStreetMap, Geonames, and Quattroshapes data into Elasticsearch, and a simple server to handle queries and auto-complete suggestions.

A major goal of the project is to make it easy to adapt the query parser and data pre-processors to the requirements of specific languages and locations. For example, users in New York City might require searchable intersections, while French-speaking users in Quebec might need a different query parser from those in France. It should be easy to edit these rules such that little programming experience is necessary.

This presentation will cover the basic architecture and data strategies of the project, provide some details into Elasticsearch's geo and completion support, and will have a big focus on how to contribute.

About the author

Randy is CEO of Mapzen, a company working on an open source mobile mapping application. In a prior role at MapQuest, he worked to re-orient the company toward open data and tools.