APAN Conference, Fukuoka, Japan
3721 Keyword Service in China
— Our thinking and Our Practice
Inter China Network Software Co.Ltd
Janurary 22th, 2003
1
Today’s Agenda
The DNS Challenge and the Keyword Solution
3721 Keyword Service and Development Status
Technical Considerations for 3721 KW Service
2
DNS technology system developed for 1980s needs
In1983, John
Postel and his
RFC #881
officially rolled
out the IETF
DNS standard
• ASCII based • 1-to-1 match • Must be exact
3
Global use of Internet poses challenges on DNS
1. Entities lose their
real world names and
identities online
2. Difficult for users of
different languages and
scripts to find info
The current DNS infrastructure results in a major gap
between online and real world names and identities,
familiar names in daily life become …?
南方周末
招商银行
http://www.nanfangdaily.com.cn/zm
http://www.cmbchina.com/
The DNS works well as identifiers for machines, but
is now increasing burdened by people around the
world who use it to search or to guess….
4
“Above DNS” navigation and search application?
“The Internet is going through an identity
crisis. DNS was never intended for
today's purposes … Higher-level locating
and searching, based on content, subject
or company names should have long ago
mooted the DNS system.“
Bob Metcalfe – Agenda 2000 Panel discussion featuring Ed
Zander, President and COO of Sun Microsystems, Mike
Capellas, President and CEO of Compaq, and Ellen Hancock,
Chairman and CEO of Exodus Communications, and Steve
Ballmer, President and CEO of Microsoft Corporation.
5
Toward a more “human friendly” Internet
“Machine Identifier” VS “Human Friendly Names”
The Future Internet Should Be
Human Friendly, Internationalized & Localized
1. How to balance the need of a technology
infrastructure that requires a good “machine
identifier” system and at the same time, easy for
human to use?
2. How to support the human friendly navigation
and search of the Internet by people of different
languages and scripts?
6
Is it just a “road sign” problem ?
Internet
IDN only alleviates some of the “pain”, but is far
from offering a truly “Human Friendly” Internet
navigation experience.
7
Chinese Keyword Enables Direct Navigation and Name
Directory Search for Online Identities
No need to remember long and complicated domain
names/URL, everyday brand and names are tools for
intuitive, direct online navigation
8
Keyword -- an application layer on top of DNS
Application
Layer
招商银行
www.cmbchina.com
192.134.1.80
“Keyword”
DNS/URL
IP
address
9
Klensin’s framework on “above DNS” applications
App Layer
on top of
DNS
DNS
DNS Layer
Layer
IP Layer
10
Ways for finding entities on the Web
Data
amount
name
or IDs
Subject, topic,
area, category
content or DB
All the info on the Web
Find an entity if you know
the name, part of the name
or its online identifier (URL)
Find an entity if you have
some general idea of which
“area” or “subject” to look for
Find an entity if you know
some terms relevant or
related to the entity
11
Frequency of usage for finding entities on the Web
KW Names or
part of the names
subject,
category,
yp dir, db
SE
“Twin Momochi Hotel”
“Hotel Twin Momochi”
“Momochi Twin Hotel”
“Twin Hotel”
Momochi area hotel list
APAN conference site
Fukuoka hotel directory
“Fukuoka”
“Hotel”
“Medium price”
12
KW is a name directory service, link names to URL
Real Life
Internet users
Online Biz
招商银行
KW
招商银行
招商银行
online
www.cmbchina.com
“Keyword service, in essence, is a name
directory service, linking real world names to
corresponding online URLs; keyword database
is a name directory database”
---- Karen Liu, 3721
13
Above-DNS KW application for human friendly web?
In our input to the US Academy of Science study project of the “Future
Internet Navigation”, 3721 envisions using KW technology to enable a
more human-friendly interface for web navigation and search;
14
Today’s Agenda
The DNS Challenge and the Keyword Solution
3721 Keyword Service and Development Status
Technical Considerations for 3721 KW Service
15
3721 KW service gaining strong usage popularity
Daily Average KW Queries
30 Mil
15 Mil
35K
Oct1999
1Mil
Oct2000
Oct2001
With over
32M unique
users per
month,
3721 is
currently
ranked #46
on a global
scale by
Alexa.com
Oct2002
16
Rapid Growth of Chinese Keyword Usage
Average daily Usage over the Weeks
周日均查询量统计
查询总量
35000000
30000000
通过软件产生的查询量
通过网站使用的次数
网站直达
特殊词查询
Usage directly in
browser address bar
25000000
20000000
15000000
Direct navigation
查询量
10000000
5000000
0
14
23
01
10
19
28
05
16
25
03
12
21
01
10
19
27
- 11 0 00 01 0 00 04 0 00 06 0 00 08 0 00 10 0 01 01 0 01 03 0 01 05 0 01 08 0 01 10 0 01 12 0 02 03 0 02 05 0 02 07 0 02 09
8
0
1
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
9 91 0 11 7- 0 32 6- 0 60 4- 0 81 3- 1 02 2- 1 23 0- 0 31 0- 0 51 9- 0 72 8- 1 00 6- 1 21 5- 0 22 3- 0 50 4- 0 71 3- 0 92 10
0
0
0
0
0
1
1
1
1
1
2
2
2
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
— Average daily usage over 30 million times
— Service available to over 99% of Internet users in China
through a combination of client software enabled browser
(~70%) and IE browser integration (~29%);
日期(周)
17
Key Features of the 3721 Keyword Service
1. Respect real world names;
— We are not in a biz of selling or assigning names;
we are in a business of providing linking service of
real world names with corresponding URL;
2. Support directory search, not just 1-to-1 look-up;
3. Support different fuzzy and user-friendly query methods;
4. Service not restricted to browser address bar;
18
3721 KW respects real world names & identities
“Refrigerator”
is a generic
word; always
results in a
directory list;
We do sell
top
placement
and direct
website
navigation,
but always
list the
directory
3721 Chinese Keyword Service follows the real world business
registration and trade mark rules; leave interpretation or arbitration of
unique “identity ownership” to the relevant Business Registration Bureau;
19
Non-unique names always result in a directory list
20
3721 offers powerful fuzzy search capability
Full name
Abbreviated
name
Wrong
word order
Misspell
Pinyin
Pinyin
initials
21
3721 KW available for use on many diff channels
Browser
SearchEng
Portal site
ISP portals
E-com sites
Mobile Internet
22
KW query integration on different channels
Browser
address
bar
Over 70% of
online PC in
China enabled
for KW service
by our client
Local
ISPs or
portals
Partnership
with nearly 200
local ISPs and
portals
Major
portals
and
Search
engines
Back-end
support for
Sina, Netease,
China.com,
Tom, Tom etc
23
Over 300K KW names for biz and entities in China
宁波市政府
招商银行
诺基亚8250
2008北京奥运
中国残疾人联
合会
中国青少年发
展基金会
24
3721 KW being proactively used by government
and business entities
Chinese Keyword
becomes Key partner of
the e-government and ehome project in China;
Eg: Dozens of Ningbo
city government
agencies use KW names
on website or print media
25
3721 KW supported by various partners
— Major portals and Search Engines
— China Telecom, CNC, and various local ISP portals
— PC manufacturers and consumer software vendors
— Various Browsers
26
Today’s Agenda
The DNS Challenge and the Keyword Solution
3721 Keyword Service and Development Status
Technical Considerations for 3721 KW Service
27
3721 primarily use client side solution
URL resolved to domain
names and DNS to IP
Internet
WWW
Keyword names resulted in 1
or many names and their URL
A KW name
Various client or
user agent
Back-end server
cluster process high
concurrent KW queries
and resolving them into
corresponding name
list and URLs
28
Major Technology Features of 3721 KW Service
• Back-end design based on XML and UNICODE:
• Support different browsers, OS, and platforms;
• Proprietary database search software technology, able to
handle high concurrent keyword query traffic and fuzzy
search; cluster servers with strong load balance and fault
tolerance capability
29
Chinese Friendly usage calls for “fuzzy search” capability
• Chinese language characteristics
— Simplified vs. Traditional Chinese input
— Words or phrases not separated by space
• Common use of abbreviated names
— No consistent name abbreviation
— Difficult to remember full name
• Difficult to Input, esp on mobile Internet
— Support Pinyin and Pinyin initial queries
These
natural, real
life use of
Chinese
language
calls for
“fuzzy
search”
capability
— Support KW use on Mobile Internet
30
Fuzzy query w/ partial match or phonetic pinyin
KW query for “information industries”
“信息产业”
— Simplified Chinese
“信息產業”
— Traditional Chinese
“Xin xi chan ye”
“xinxichanye”
“xxcy”
— Phonetic Spelling (pinyin)
— Phonetic spelling initials
31
Fuzzy query using abbreviated Chinese names
KW query for
“Beijing 2nd Foreign Language Institute”
“北京|第二|外国语|学院”
“北京第二外国语学院”
“北二外” — (北京第二外国语学院)
“北京外国语学院” — (北京第二外国语学院)
32
UNICOE, XML based expandable to other language
Support foreign characters, numbers, homonyms ..
“Ericsson”
“500028”
— English Alphabet
— A Chinese Stock Ticker
“Nokia8250”
— Combination
“武夷山” (wu3 yi2 shan1)
“五一山” (wu3 yi1 shan1)
— Homonym
33
3721 actively develops KW app in wireless Internet
34
Make effort to participate in Intl Tech Std Dev’t
Submitted two
Internet Drafts
(I-D) to IETF:
1. “Caching Mechanism in Layered DNS search System”
http://www.ietf.org/internet-drafts/draft-xhshi-dns-search-caching-00.txt
2. “Integrating Layered DNS Search Service within User Agent”
http://www.ietf.org/internet-drafts/draft-xhshi-dns-search-00.txt
35
Thank You!
Inter China Network Software Co. Ltd
Karen Liu
[email protected]
36
Descargar

Internet Conference - QGPOP (Kyushu GigaPOP Project)