How the New Japan Geocoder Can Help You

We are pleased to announce a new Japan geocoder that improves how you search for a Japanese address on http://www.bing.com/maps and how you geocode a Japanese address using the Bing Maps APIs. In this post, we will introduce you to the new features of this geocoder and show you how to use these new features with the Bing Maps REST Services API. We’ve received more than a few inquiries about Japanese address geocoding, so we decided to write a detailed post.

You can also find details about Japanese geocoding using the Bing Maps APIs on MSDN.

Let’s start with some background.

The Bing Maps geocoder is a global geocoding service, but there are many system components that are responsible for geocoding individual markets. One such component is the Japan geocoder, which resolves Japan in-market queries. The Japan geocoder is rooftop based, and relies on Zenrin data provider address points. Zenrin has phenomenal coverage over Japan (99.6%), with the exception of a few offshore islands. Statistically, there’s about a total of 36.1 million rooftop address points that are geocodable. Also, we use more than 250,000 geocodable places of interests.

Japanese address geocoding is complicated because there are three different types of native character sets (Kanji, Hiragana, and Katakana). Also, there is a fourth non-native character set that is the romanization of Japanese using Latin script (Romaji). Let’s take a look at how the new geocoder features support these character sets and more.

New Japan Geocoder Features

Below we will introduce major features of the new geocoder and show how they solve problems users have faced. These features include support for:

  • Latin characters
  • Western and Japanese address systems
  • Mixed Latin and Kana characters
  • Kana variants

Western and Japanese Address System Support

There are times when you want to geocode a Japanese address that does not use the convention for addresses in Japan. In the past, there was a geocoding issue when a user would enter a Japanese address based on the western address convention (see below for explanation). This was often the case when the address was written in Latin characters. Also, the order of items in the address could be chaotically jumbled because of lack of understanding or some data store issue.

Japan Geocoder

 

To illustrate, if you have ever sent a letter to Japan using the western addressing system, what you write on the front of the envelope is typically in Latin characters (Romaji):

“1-12-1 Nihonbashi Chuo, Tokyo, 103-8260, Japan”

If you translate this into Japanese characters, it looks like this:

“1-12-1,日本橋,中央区,東京103-8260,日本”

However, the Japanese convention is to write this address in reverse order:

“〒103-8260東京都日本橋中央区1-12-1”

Note:In the Japanese convention, the country Japan (日本) is not required and symbols for postcode (〒) and prefecture (都) are required.

A typical Japanese address written using the Japanese addressing system uses the opposite convention of addresses in western countries such as the United States and Europe. Below is a comparison of the Japanese and western address conventions:

Japanese Addressing System

(postcode) (prefecture) (city/municipality/ward) (location in city/municipality/ward) (detailed sections)

Example: {108-0075 東京都 港区 港南2-16-3} which reads as {108-0075 Tokyo-to Minato-ku Konan 2-16-3

Western Addressing System

(detailed sections)(location in city/municipality/ward) (city/municipal/ward)(prefecture)(postcode){country}

Example: “2-16-3 Kounan Minato-ku, Tokyo 108-0075 Japan”

The new Japanese geocoder is far more robust than its predecessor and can now geocode addresses that are not in the correct order. We are no longer requiring that the address use the Japanese addressing system.

Latin Character Support

Latin-based Japanese (Romaji) is quite complex. Often users outside of Japan don’t know how to write Japanese using Latin characters. If you want to learn more, James Curtis Hepburn invented the Hepburn Romanization system for writing the Japanese language using the Latin alphabet.

In many cases, users and applications expect Hepburn (also known as Hebon) spelling and its variants to geocode successfully when they enter a Japanese address such as the following example:

“2-16-3 Kounan Minato-ku, Tokyo-to 108-0075 Japan”

In the example above, there are suffixes like “-ku”, “-to”. These are suffixes that come from the phonetic sounds that mean ward and prefecture respectively. In the past, the geocoder would have difficulty geocoding this Hebon version of the address. However, the new Japan geocoder can handle this type of Latin address.

The new geocoder can also handle multiple forms of Japanese Hebon spelling (e.g., “Konan” instead of the Japanese phonetic spelling “Kounan”).

“2-16-3 Konan, Minato Ward, Tokyo-to, Japan”

The new geocoder can also handle a special unicode characters that are part of Hebon such as a Latin character “o” with macron Å:

“2-16-3 Kōnan, Minato-ku, Tōkyō, Japan”

The geocoder can also handle a complex query containing both Latin and Japanese characters:

“東京都 Minato-ku Kounan 2-16-3”

Note that a Japanese address written in Latin form (Romaji) typically uses the western address convention that is reversed from the Japanese addressing system. The new geocoder will handle either system.

In the past, there was also confusion about whether you needed to specify “Japan” as the country in your geocode request. Before, you needed to specify “Japan” unless the address was written in Japanese. The new geocoder clears up this confusion by not requiring “Japan” for any form of Japanese addresses. If the geocoder identifies the address as a Japanese address, it will handle it.

Note that when you make a geocode request using the Bing Maps REST Services or Bing Spatial Data Services, you do not need to specify “Japan” in your address, but you will need to set the culture parameter/value to “ja”.

All in all, the new Japan geocoder is able to handle Latin and is less susceptible to variants arising from using different spellings or notations.

Additional Japanese Kana Variants Support

You may wish to create queries that contain phonetic Hiragana or Katakana representations instead of the logographic systemof Kanji as shown in the following examples:

“ながのけんすわし”  (Hiragana)

“ナガノケンスワシ”  (Katakana)

The English translation of this query is “Nagano Prefecture, Suwa city.”

Previous versions of the Japan geocoder have not been able to handle these type of queries. However, the new geocoder can recognize these representations as the Japanese Kanji official address:

長野県諏訪市

The new geocoder can also interpret partial variants that contain a mixture of Kanji, Hiragana, and Katakana:

長野県オカヤ市

Typically, this ability to resolve Japanese Kana instead of Kanji is extremely useful when the user remembers the sound of the address, but not the Kanji representation. While we do not guarantee that we can resolve the address for every case, the new geocoder is designed to be better at handling these cases.

Custom / Additional Address Support

The new Japan geocoder has the additional ability to handle custom addresses or unofficial addresses. For example, if an address contains additional information that is not part of the official address, you may still be able to geocode the address.

Address alone: “宮崎県日向市日知屋1383-51”

Address with extra value: “宮崎県日向市大字日知屋1383-51”

While we do not guarantee that we can geocode every possible unofficial address that arises, the new geocoder does have the framework to handle these situations.

Examples Using the Locations (Geocoding) API [Bing Maps REST Services]

Below are some examples of geocoding locations in Japan using the Locations (geocoding) APIthat is part of the suite of Bing Maps REST Services.

Note: In the examples below, you can replace the Japanese Kana with Latin characters.

Geocoding Using the Locations API Address Parameters

Geocoding using the URL address parameters should only be used when the user knows the address hierarchy mapping. Typical address mappings for Japan to the Locations API Find by Query parameters are as follows:

Used Fields Examples
CountryRegion {JP} :
PostalCode {〒168-0063} : the symbol {〒} is optional.
AdminDistrict {東京都} : in the case of Tokyo prefecture{北海道} : in the case of Hokkaido prefecture{大阪府} : in the case of Osaka prefecture

 

{長野県} : in the case of all other prefectures

{日本} : if you want to include the country name, Japan, append in front, such as {日本、東京都}.

Locality {杉並区} : in Tokyo or any other special designated cities{諏訪郡} : in prefectures with gun{武蔵野市} : similar in case of all other municipals
AddressLine {和泉3-61-43} : all address subsections below municipals

Geocoding Using the Locations API and a Single Query Address String

You can geocode by specifying a string that contains address data. Unlike using URL address parameters, you do not need to identify the different address values. For more information about what an address string would look like, refer to the address conventions described in the previous section on western and Japanese Address Systems and the mapping table above. For queries using a single address string, it is best to specify the address in the western or Japanese order as described in the western and Japanese Address Systems Support section of this post.

REST URL Examples

Note please replace the placeholder “xxx” with your Bing Maps Key in the queries below. The Latin character and Japanese Kana query examples will also work if you reverse the order of the address values. Note that you must set the culture (c) parameter to “ja” for these URLs for the geocoding to work correctly.

Category Full Address OR corresponding address sections, coordinates URL
full address using URL parameters PostalCode = 〒168-0063

 

AdminDistrict = 東京都Locality = 杉並区

AddressLine = 和泉3-61-43

http://dev.virtualearth.net/REST/v1/Locations?
countryRegion=JP&adminDistrict
=%E6%9D%B1%E4%BA%AC%E9%
83%BD&locality=%e6%b8%af%
e5%8c%ba&addressLine=%e6%
b8%af%e5%8d%97%ef%bc%92%
e2%88%92%ef%bc%91%ef%bc%
96%e2%88%92%ef%bc%93&o=xml&
key=Xxx
Full address using query string Query = 〒168-0063 東京都杉並区和泉3-61-43 http://dev.virtualearth.net/REST/v1/
Locations?query=%e3%80%92108%2
d0075%e6%9d%b1%e4%ba%ac%e9
%83%bd%e6%b8%af%e5%8c%ba%e
6%b8%af%e5%8d%97%ef%bc%92%e
2%88%92%ef%bc%91%ef%bc%96%e
2%88%92%ef%bc%93&o=xml&c=ja&
key=Xxx
postcode geocode using postal code parameter PostalCode = 108-0075 http://dev.virtualearth.net/REST/v1/
Locations?countryRegion=JP&postal
Code=108-
0075&o=xml&key=Xxx&c=ja
postcode geocode using query string Query = 〒168-0063 http://dev.virtualearth.net/REST/v1/
Locations?countryRegion=JP&postal
Code=%e3%80%92108%2d0075&o=
xml&key=Xxx&c=ja
Reverse Geocoding Latitude=35Longitude=139 http://dev.virtualearth.net/REST/v1/L
ocations/35,139?o=xml&key=xxx

Important Note:

For URLs containing Japanese Kana, please base64 encode the utf8 strings.

For reverse geocoding, responses are returned in Japanese native character sets only (Kanji, Katakana, or Hiragana).

Final Words

Japan geocoding is complex because, linguistically, it must handle four types of characters (Kanji, Katakana, Hiragana, and Romaji). Three of these types of characters are native to the language (Kanji, Katakana, and Hiragana) while the fourth uses Latin (Romaji) ) characters. Romaji is not commonly used by Japanese native users for addresses. However, its usage has become more common, especially for clients outside of Japan who are interested in geocoding addresses in Japan, but cannot read Japanese.

Also, Japanese addresses do not contain street names. (There are exceptions like Kyoto Toorina addresses). Japanese addresses are generally based on blocks with representative centroids.

Japanese geocoding is not difficult if you understand the Japanese addressing system and how to map Japanese address values to the Locations API URL address parameters. If you are not comfortable with the address parameters, you can stick to the query string method. The new features of the Japanese geocoder that allow for addresses to be specified out-of-order and that accept Latin and other Japanese character sets should help you with your Japanese geocoding needs.

– Tetsuaki Otsuki, Program Manager, Bing Maps Japan