Yahoo! GeoPlanet Exploring Places (without Maps)

1
Yahoo! GeoPlanetExploring Places (without Maps)
2
Intro

• Yahoo! Geo Technologies overview
• The issues in and around global geo data
• GeoPlanet Web Service

3
Yahoo! Geo Technologies

• Expertise
• Maps
• global mapping platform and geospatial
  technologies
• User Location
• technologies and platforms for detecting and
  managing user location
• Geo Informatics
• global ontology of named places, geoparsing
  geo-enrichment
• Mission
• Collect the worlds geography as it is used by
  the worlds people
• Make Location a core component of the entire
  Yahoo! experience
• Connect the web world with the real world
• Open up Geo Platforms

4
Geo-Parsing

• Geo Parsing
• Recognition, Identification, Disambiguation and
  Retrieval of named places in text
• Sounds easy, doesnt it?

5
Geo Parsing the problems

• Is it a place or not a place?
• Person Joe Orlando, Paris Hilton, Jack London
• Title Miss Ohio
• Movie Letters From Iwo Jima, Gran Torino
• Song 24 Hours From Tulsa
• Company Products Waterford Crystal, Bryn Mawr
  Stereo
• Cars Chevy Malibu, Chevrolet Corsica
• Clothing Panama hat, Oxford shoes
• Food Yorkshire Pudding
• TV LA Law, CSI Miami, CSI New York, Chicago Hope
• Magazine ???? (Doda)

6
Geo Parsing more problems

• OK, so what place do you mean?
• Colorado, Texas
• Oregon, Alaska
• Peru, Bolivia
• Peru, Argentina
• Washington, Nevada County, California
• Kansas, Delaware County, Oklahoma
• East Texas, Pennsylvania
• East London, South Africa
• York, New York

7
Springfielddoh!
Springfield, Illinois Springfield,
Massachusetts? Springfield, Missouri? Springfield,
Vermont? Springfield, Oregon? One of the other
77 Springfields in the US? Or one of the 35 other
Springfields outside the US?
8
Things that are sent to test us
9
Overview

• Yahoo! GeoPlanet is a publicly available web
  service for managing named places around the
  globe in a consistent, language-neutral way. It
  provides the geographic developer community with
  the vocabulary and grammar to traverse the
  world's geography, facilitating the
  geo-enablement of applications relevant to users
  and their businesses, by providing a means to
  recognize, locate and exploit geo-relevant data
  across the Web
• http//where.yahooapis.com
• Youll need a valid YDN app ID to use GeoPlanet

10
GeoPlanet in a Nutshell

• REST API for geographic discovery and mining
• Resource of over 6m named places in 21 languages
• Permanent Identifiers
• Global Coverage, Consistent Framework
• Integrates with Flickr, FireEagle
• Answers questions such as
• What zipcodes are in the city of Chicago?
• How many Springfields are in the US?
• Which is most likely to be the Springfield I
  mean?
• What is the unique ID (WOEID) for The Sydney
  Opera House

11
the WOEID

• The Where on Earth identifier
• assigned to every named place
• 32-bit
• Unique
• Permanent
• Language neutral
• London Londres Londra ????
• Non-Repetitive
• Never Recycled
• Deprecations mapped to successor/replacement
• Requests to the service using a deprecated WOEID
  are served transparently.

12
WOEIDs WhereHaus Global Geographic
Ontology
23689940
23511738
13
The GeoPlanet Hierarchy

• Continents
• Countries
• Counties
• Regions
• Area Codes
• Marketing
• Neighborhoods
• Zipcodes
• Custom

13
14
The Thinking Place-Based Logic
15
The Resource WhereHaus

• WhereHaus
• Global Locations Database
• place names
• post codes
• administrative regions
• telephone code areas
• marketing areas
• points of interest
• colloquial areas
• neighbourhoods

16
Resources and
Collections

• Information in GeoPlanet is organized into
  collections and resources
• Resources are unique objects that contain
  multiple attributes
• Resources use singular names (place)
• Collections are lists of related resources
• Collections use plural names (places)

17
Place Resource

• Usage
• Get information about a place
• Template
• /v1/place/woeid
• Examples
• /v1/place/1 Get information about the World
• /v1/place/23424977 Get information about the
  USA
• /v1/place/2487956 Get information about San
  Francisco

18
Place
Relationship Collections

• Usage
• Get information about places related to a place
• Template
• /v1/place/woeid/relation
• Relations
• Parent 1st degree containing entity
• Children 1st degree contained entities
• Neighbors adjacent peer place types
• Siblings peer place types sharing same parent
• Belongtos collection of containing entities
• Ancestors higher administrative hierarchy

19
Some Examples

• /v1/place/1/children.type(Country)
• Get list of countries
• /v1/place/2487956/parent
• Get parent place of San Francisco
• /v1/place/2487956/belongtos.type(TimeZone)
• Get Olson timezone name for San Francisco

20
Some more
Examples

• /v1/place/2347563/siblingscount0
• Get list of states (siblings of California)
• /v1/place/23424977/children.type(State)
• Get list of states (state children of United
  States)
• /v1/place/12797497/neighbors
• Get list of zip code adjacent to 94089

21
Places Collection

• Usage
• Find places that match a place name, and
  optionally a place type
• Templates
• /v1/places.q(placename,focus)countcount
  startstart
• placename must be enclosed in single quotes if
  it contains a comma

22
Places Collection

• Usage
• Find places that match a place name, and
  optionally a place type
• Templates
• /v1/placesand(.q(placename,focus),.type(pl
  acetypes))countcountstartstart
• placetypes is a list of up to seven placetypes
  each placetype can be either a place type id
  (e.g.7) or place type name (e.g. Town)

23
Place Type Names
and Codes

• http//where.yahooapis.com/v1/placetypes
• Some examples
• Continent 29
• Country 12
• Admin1 8
• Admin2 9
• Admin3 10
• Town 7
• Suburb 22
• Postal Code 11
• Supername 19
• Colloquial 24
• Time Zone 31

24
Some Examples

• /v1/places.q(Springfield) find most likely
  place named Springfield
• /v1/places.q(Springfield)count0 find all
  places named Springfield
• /v1/places.q(Rome,Italy) find Rome, Italy
  (note quotes around place name)
• /v1/places.q(Roma,Italia) find Rome, Italy
  (using Italian place name)
• /v1/placesand(.q(Soho),.type(Suburb))count0
  find suburbs named Soho

25
Other Concepts

• Representation
• - The short - contains only the most essential
  information
• - The long - includes additional elements
• Language
• RFC4646
• Positional Consistency
• Centroid and Bounding Box
• Longitude/Latitude to WGS84 datum
• Response Formats
• xmlXML
• jsonJSON
• geojsonGeoJSON

26
References

• YDN landing page http//developer.yahoo.com/geo
• Documentation http//developer.yahoo.com/geo/gui
  de
• YDN Developer Reg http//developer.yahoo.com/ysr
  egapp
• GeoPlanet Forum - http//developer.yahoo.net/forum
  /index.php?showforum31
• Yahoo! Geo Technologies Blog - http//www.ygeoblog
  .com

27
Questions

• Martin Barnes Data Manager Yahoo! Geo
  Technologies UK
• barnes_at_yahoo-inc.com
• Gary Gale Head of UK Engineering - Yahoo! Geo
  Technologies
• garyg_at_
• Walter Andrag Lead Engineer Yahoo! Geo
  Technologies UK
• walter_at_
• Roman Kirillov Senior Engineer Yahoo! Geo
  Technologies UK
• kirillov_at_
• Lots of knowledgeable YDN people (S. Major
  April 09)

28
and Finally. The future.

• We will be at Where2.0 in San Jose later this
  month
• http//en.oreilly.com/where2009/
• Watch out for more news from Geo Technologies!

29
Happy Hacking
Hack Day Haiku Give me your content With love, I
will geoparse And return places