Our Pick of

The Top Japanese
Search Engines

by Tina Lieu

Back in the early days of the Internet, you probably picked your search engine based on how many matches it found in response to your query. That was the case in Japan even just a year ago, when Japanese-language content on the Internet was still sparse.

Today, though, we're drowning in data, and quality is more important than quantity. Rather than slog through 1,000 results, you want your search engine to narrow the choices down to a handful of quality sites that offer the information you need.

Nowadays, any self-respecting search engine will allow you to input at least two keywords, with your choice of Boolean operators ("OR" and "AND") to indicate whether or not each page in the search result should contain either or both of the keywords. But there are several variations on this theme.

A key point to keep in mind when using a Japanese search engine is the need to use single-byte spaces (hankaku space) between keywords, not double-byte spaces. Also, none of the search engines surveyed for this article permit hankaku katakana (hankaku katakana) input.

To evaluate Japanese search engines and their search-refinement techniques, we searched for information on famous Japanese film director Juzo Itami. (Itami is also the name of a city in Japan, and the characters for "Juzo" have a more common pronunciation of juusan ("thirteen").) Except where indicated, we used kanji input for the searches - this is the recommended input for most sites. Many search engines will accept romaji and katakana input, but unless the search term appears in katakana or romaji in the text of a webpage, such input will not bring up a match.

The search results turned up by most engines include either a brief summary or a few lines of text from the located document, plus the webpage's title and URL. Note that some search engines have a separate URL for "advanced" searches. These are marked "(refined)" in the following search engine descriptions.

The information and URLs in this article were current as of late September.

ACARANAVI
http://webdew.rnet.or.jp http://webdew.rnet.or.jp/service/shank/NAVI/SEARCH/search2.html (refined)

ACARANAVI posts the number one accessed link from the previous day.

Search refinement - Single-byte spaces between keywords can be set as "AND" or "OR." There is a check box for returning only sites suitable for minors (age 18 and under). The search results can be further refined by selecting a search category or entering an additional keyword into a blank that appears at the bottom of the search result list.

Search results - Search results are listed with the category each page falls under, the page's percentage of relevancy to the search, and whether the page is oriented towards people who want to "look" or "learn/know".

GOO
http://www.goo.ne.jp

GOO (developed by NTT and Inktomi) opened in March 1997, and it really does seem to have the muscle claimed at its opening. GOO is powered by Inktomi and InfoBee, and is associated with G Square.

Search refinement - GOO offers a full range of search-refinement options. You can search Japanese-language sites or overseas sites, but not both simultaneously. You can choose one of six search refiners: include all terms, find any one term, use Boolean operators, search for pages of all languages, rank results by language, or rank results by URL (no summary). Advanced features allow restricting the search results with additional keywords, or by date (created or updated), data file type, or server domain.

Search results - The results show each page's percentage of relevancy to the search, URL, date, and file size (number of bytes). The huge number of returns on our trial search (see the accompanying table) shows the large number of sites GOO encompasses. The results, however, seemed good, and very good after adding the second term into the search.

Hole-in-One
http://hole-in-one.com
http://hole-in-one.com/search.html (refined)

Hole-in-One, which has been around for a while, is sponsored by Hitachi International Business.

Search refinement - You have a choice of using "AND" or "OR" between search terms. You can select to search the Japanese Web, or only Hole-in-One (presumably registered sites). If you select Japanese Web, you can choose to look for your search terms within page contents, title, or URL. If you select Hole-in-One, you can choose one of three search locations: (1) title, description, keyword, (2) category, or (3) URL.

Search results - The search results come with links to their appropriate categories so that you can look for related sites.

InfoBee
http://navi.ntt.co.jp/index.html (English)
http://navi.ntt.co.jp/home.html (Japanese)

This site, which covered 176,159 pages as of late September, is the same as the NTT Directory. The Titan search engine is also accessible from here. The Japanese and English search engines are identical; only the language of the interface is different.

Search refinement - You have a choice of "AND" or "OR" between search terms. Kanji input is recommended.

Search results - After receiving the search results, you have the option of adding another keyword to further refine the list.

InfoNavigator
http://infonavi.infoweb.or.jp

Search refinement - InfoNavigator accepts the Boolean operators "AND," "OR," "NOT, and "ADJ." ("ADJ" specifies that the second search term appears on the page close to the first.) You can also choose to search for time-sensitive sites, such as those running limited-time contests/campaigns or those associated with festivals.

Search results - The URLs of the search results are not displayed, and the comments are those that were written by the website submitters. The results are ranked by relevancy or date entered into the InfoNavigator database.

Infoseek
http://japan.infoseek.com

This search engine accepts only submitted websites that contain some Japanese text. For Windows 95 users, Infoseek offers a small downloadable program, called "QuickSeek," that enables you to embed the Infoseek search engine input line into your MS Internet Explorer 3.0 or Netscape Navigator browser for use in searching either Infoseek Japan or Infoseek US (http://www.infoseek.com).

Search refinement - InfoSeek offers a huge menu of search-refinement options. Even "natural language" phrases can be input, and will sometimes turn up good results. You can also search for a term within a specified domain (such as "kenkyuukaihatsu"- within only ac.jp domains).

Search results - You have the option of refining the list by adding another keyword. The percentage of relevancy and file size of each match are given. The search results for our trial were generally good.

ODIN
http://kichijiro.c.u-tokyo.ac.jp/odin/

ODIN stands for Open Documentary Information Navigator. The search engine robot abides by certain "rules" that webmasters can use to avoid having specified pages catalogued. The search is not case sensitive for terms, although it will differentiate between all-uppercase terms and those containing some lowercase letters. Searching for "HTML," for example, will not find pages with "Html" or "html."

Search refinement - ODIN has several search refining techniques. A "+" before a term requires that it appear. A ">" can be used as in "A>B" - which means that "out of the pages containing A, find the pages that also contain B." An "*" represents a wild card character. Search results - The results are ranked by relevancy.

RCAAU Mondou
http://www.kuamp.kyoto-u.ac.jp/labs/infocom/mondou/search.html (Japanese)
http://www.kuamp.kyoto-u.ac.jp/labs/infocom/mondou/index_e.html (English)

The acronym in the name stands for "Retrieval loCation by weighted AssociAtion rUle." This search engine is especially nice because it uses an associative method: after searching for one keyword, several related words will be suggested for use in narrowing your search. Instructions are in both English and Japanese.

Search refinement - The romaji input is case insensitive. Hiragana is not accepted, only kanji or romaji. The user can select terms to include or exclude, or choose to find exact matches or words that begin with a specified term.

Search results - The associated terms offered up when we did our search for iami and then juuzou were quite surprised in their accuracy. The search for itami alone brought up itoh, juuzou, kan, Kumamoto, shi, Osaka, itami kuukou and shiritsu. Clicking on juuzou brought up super, uper, seki, eiga, onna, suupaa, and kantoku. Each term in the second search is closely related to the films of Juzo Itami or to the director himself (except for seki which appears as in kansuru "about.")

Senrigan
http://senrigan.ascii.co.jp (Japanese)
http://senrigan.ascii.co.jp/index-e.html (English)

Senrigan limits its database to pages in the .jp domain; overseas Japanese-language pages will not appear. Senrigan claims to have a database of 1.27 million URLs.

Search refinement - Users can select what kind of files to return: all, ftp, http, or mailto. Spaces between words are equivalent to "AND," but no other Boolean operators are accepted. Search results can be sorted by fewest broken links.

Search results - The Search results indicate how many broken links are on each page. This may (or may not) be an indication of how well the page is being maintained - it could be just poor HTML encoding.

Yahoo! Japan
http://www.yahoo.co.jp

This site is associated with the well-known English Yahoo!

Search refinement - Yahoo allows "AND " or "OR" to combine multiple search terms. Its "expert search" is quite comprehensive. " * " between two terms means "AND"; "+" means "OR"; "#" means "NOT"; "(A), B" means "find all the pages that contain term A, then look for term B"; and "?" is the wild card for either matching the "head" (A?) or "tail" (?A) of terms.

Search results - The Search results show where the keyword appears in the webpage. If the keyword appears as a category name, it will pull out the category.

Search Engine	URL	Itami	Itami and Juuzou	Features
* ACARANAVI (refined)	http://webdew.rnet.or.jp/service/shank/NAVI/SEARCH/search2.html	19	3	S,C, R
CSJ Index	http://www.iijnet.or.jp/csj/	25	0	S,C
Dragon	http://www.dragon.co.jp/	15	3	S,C
Excite (Japanese)	http://jp.excite.com	(itami) 583	(Itami, Juuzou) 617	S
* GOO (Japanese)	http://www.goo.ne.jp	6414	612	S, C, R
* Hole-in-One (searching Japanese Web)	http://hole-in-one.com/	323	36	S,C
* Hole-in-One (searching Hole-in-One)	http://hole-in-one.com/	15	3
* Infobee (Japanese)	http://navi.ntt.co.jp/home.html	28	5	S,C
* InfoNavigator	http://infonavi.infoweb.or.jp/	32	4	S,C, R
* Infoseek	http://japan.infoseek.com	5440	186	S,C, R
			(Itami, Juuzou) 355
Netplaza	http://netplaza.biglobe.or.jp/index.html	20	4	C
Nikkei Net	http://www.nikkei.co.jp/ss/	5	1	S,C
Nippon Search Engine	http://www.juno.sfc.keio.ac.jp/NSE-NS/dive/	6	1	C
* Odin	http://kichijiro.c.u-tokyo.ac.jp/odin/	411	(Itami, Juuzou) 37	R
* RCAAU Mondou 1	http://www.kuamp.kyoto-u.ac.jp/labs/infocom/mondou/search.html	225	16	S, R
* Senrigan	http://senrigan.ascii.co.jp/	72	7	R
Titan (English)	http://isserv.tas.ntt.jp/chisho/titan-e.html	40	(Itami, Juuzou) 40	S,C
Titan (Japanese)	http://isserv.tas.ntt.jp/chisho/titan.html	70	0
URL Square (ORIONS)	http://www.orions.ad.jp/urls/index-jp.html	7	1	S
Wave Search	http://www1.sony.co.jp/InfoPlaza/WAVESearch/	53	9	S
Yahoo	http://yahoo.ita.tutkie.tut.ac.jp/yahoo/search.html	14	0	S,C
* Yahoo Japan	http://www.yahoo.co.jp	49	5	S,C
Search results looking first for Itami (column 1), then for two terms Itami and Juuzou (column 2).
* Where the terms used for the search differed from those given in column heading, the words and/or search refinement symbols are listed in parentheses.
* The ten " tier two" search engines (those not marked with an asterisk in the table) are not covered in this article. However, evaluations of these search engines are included in the extended version of this article in the Online Extras section of the Computing Japan website, http://www.computingjapan.com.
* "S" (submit) indicates that the search engine accepts webpage submissions. "R" (robot) indicates that the engine uses an automated robot (aka Web crawler) to locate and check out sites. "C" (catalog) means that the site catalogs website pages by topic for easy browsing.