ABSTRACT
This paper provides the international WWW community with a pre-release glimpse of innovations pioneered by the Japan Window project in information content and technical features. Japan Window is a US-Japan university-industry research project which investigates the following issues: human-computer interfaces (HCI), Japan-information content standards, multi-lingual computing, and Japanese technology management research.
Japan Window's principal objective is to grow an integrated Web site useful to US corporate, academic and government organizations in their daily US-Japan activities. This is achieved by concurrently developing 1) a new and timely Japanese technology and science, government and business/economics information source for the US high technology community, and 2) cutting-edge human-computer interface features and techniques. US-based organizations are Japan Window's primary target audience. Once fully developed, Japan Window could serve as a multi-lingual model for Asia-information exchange in many areas of mutual interest.
Japan Window is a collaboration of Stanford University's US-Japan Technology Management Center (http://fuji.stanford.edu) and Nippon Telegraph and Telephone (NTT) (http://www.ntt.jp). Japan Window is an official project of Smart Valley Inc. (http://www.svi.org), and is supported in part by WAIS, Inc. *
_________
* WAIS Inc. is supporting the Stanford US-Japan Technology Management Center's work on theJapan Window project.
1.0 Introduction and History
The Japan Window project is a WWW research collaboration of Stanford University's
US-Japan Technology Management Center and Nippon Telegraph and Telephone Corporation (NTT), under the sponsorship of Smart Valley, Inc.. Japan Window is an experiment in organizing and building country level information on the Web, and it relies extensively on experience gained in the first-generation Stanford "X-Guide to Japan Information Resources" [Lee, et al, '94] and NTT "Japan Information" Web sites created in 1993-94. This combined team brings together the two best known Web sites for Japan information today to improve Japan information access and US-Japan exchange and communication.
2.0 Project Objectives
Japan Window's principal objective is to grow an integrated Web site useful to US corporate, academic and government organizations in their daily US-Japan activities, be it hi-tech R&D, product development, electronic commerce, export promotion, or technology management and policy. This is achieved by concurrently developing 1) unique and innovative human/computer interface features and techniques, and 2) new and timely science, technology and business information sources in Japan for the US high technology community. Once fully developed, Japan Window could serve as a multi-lingual model for US-Asia information exchange in many areas of mutual interest.
3.0 Target User Community
US-based organizations are Japan Window's primary target audience. Major US user communities targeted include high technology industry, particularly in Silicon Valley; researchers at government, corporate and university laboratories; and US government agencies. A planned User Committee will ensure that Web site users' feedback is systematically incorporated into the site design and testing. Japanese organizations are also expected to be heavy users of the Web site.
4.0 Research Objectives
Japan Window seeks to investigate the following issues of particular interest to the US-Japan and WWW/Internet communities:
Human-Computer Interfaces. Human-computer interface research objectives are aimed at developing new user features to 1) improve Web site utility and efficiency per client visit, and 2) maintain user interest over the long term. These features include database and Web site search and retrieval, cartographic map-based navigation, bi-lingual information display, database update/display for different user groups, interactive forms, secure pages, daily utilities and updates (such as Yen rate and graphics swapping) ("dynamic pages"), user-specific features, and shopping/electronic commerce. Custom-designed features which achieve these objectives will be demonstrated live at the conference.
Information Content Standards. The project seeks to develop standards and benchmarks for the dissemination of publicly-available information by Japanese organizations to US and other users. Our use of "homepage templates" will be pursued in an effort to encourage Japanese organizations (particularly companies and government ministries, agencies and related organizations) to provide information in important content areas, user-friendly formats and access methods suitable for US and other non-Japanese speaking users. Early cooperation with leading Japanese organizations during their Web site design is essential to the success of this initiative.
Multi-lingual Computing. The integration and maintenance of large bi-lingual information databases is an important area of on-going investigation. American users and companies also need to be educated in the use, dissemination and benefits of multi-lingual information. NTT's main Web site (http://www.ntt.jp) is bi-lingual, and provides an experiential basis for expanded US applications here.
US Market Research. Relatively little is known about the US market for Internet-based dissemination of Japan information. The project will collect detailed user statistics, and analyze these on a regular basis to determine patterns of usage to assist in on-going site design and expansion. Certain statistics will be furnished to information providers in Japan.
Japanese Technology Management Research. The process of designing and maintaining this dynamic Web site provides an excellent platform for the investigation of important Japanese technology management research issues: the impact of the Internet on technology management practices in Japanese government and industry; information, computing and networking usage patterns in Japanese organizations; and Japanese national information infrastructure (NII) management and development.
5.0 Information Content Approach
Japan Window aims to develop the US market for Japan information in specific selected domains. To date, Japanese government and industry efforts to provide Japan-information for the US market have met with generally unenthusiastic responses from American users. We believe that this is due to:
Content Strategy. The Japan Window information content strategy - and technical presentation - is, therefore, structured so as to address these problems by investigating alternative approaches directly with US users. We aim to attract the casual browser of Japan-information, as well as the more serious, repeat user who frequently needs in-depth information in one or more subject areas.
Our content strategy mixes daily dynamic information with relatively static information, so as to appeal to a broad base of users with varying Japan-information needs. The casual user will be more likely to access Japan Window for the daily updates; once at the site, he/she might be inclined to check out the government or other subject area sections.
As presently conceived, Japan Window information content will focus in the following major information categories and features:
Information Subject Areas
Duplication with Stanford's X-Guide and the NTT Homepage will be kept to a minimum through appropriate coordination and communication.
Japanese Government. English-language Japanese government information will be the first major subject area released by Japan Window. The government collection covers not only central government information but also local government information, including selected prefecture pages and city pages. In each area, users can find background descriptions of various government organizations along with its current activities and policies. To encourage these regional activities further, we are planning to identify sister city relationships between US and Japan.
Japanese Science and Technology. Japanese science and technology information, in English and Japanese, will be an important Japan Window subject area. Domain and discipline areas currently under consideration include Japanese manufacturing, environmental and energy technologies, biotechnology, and telecommunications and computer networking R&D information. It is our hope that US and Japanese governments and companies will support and use this section as we seek to develop and open up the exchange of Japanese technical information.
Japanese Business. Economics, and Finance. Japanese business, economics and financial information is a third important subject area currently under development. Ideas under discussion include daily news features, economic forecasts and performance reports, and Japanese company information. It is intended to be a useful information source for persons interested in finding out how to invest in Tokyo stocks or how to set up a computer software firm in the Western region of Kansai, for example. We hope our research will build a foundation for US-Japan international trade and commerce on the Internet.
Travel, Living and Working in Japan. Travel, living and working in Japan information will include practical and useful information for Americans who plan to go to Japan. In addition to general information, Japan Window will provide detailed maps and guides to important regions in Japan. Whenever practical, we hope to assist users with the preparation of overseas travel by indicating how to fill out visa forms, and get around once landed in Japan.
Kids Window. Kid's Window is designed for children in America to learn about Japanese culture, language, and other children with whom they can be friends. Many visual images, such as illustrations and photographs, are used with visual effects to attract children and keep them interested. By browsing this window, children will be exposed to a world very different from their own. As the children explore, they may discover that children in Japan share similar interests, dreams, etc. The purpose of Kid's Window is to serve as a communication bridge between children in America and the children in Japan. It is expected to promote understanding and appreciation of each other.
Daily Updates. We have developed an efficient mechanism to receive news feeds from various organizations and parse them to the appropriate locations on our Web site. This feature will assure Japan Window users current and timely information from original sources.
Selected Silicon Valley Home Pages in Japanese. Japan Window's multi-lingual site research will eventually be transferred to selected interested organizations in Silicon Valley. Working with Smart Valley, Japan Window will work to provide assistance in developing home pages in Japanese. One facet of this endeavor will be to examine current machine translation technologies and experiment in developing practical interfaces to the World Wide Web.
Bi-lingual Information. Japan Window will be bi-lingual (English-Japanese) whenever practical. Although the primary focus "audience" of Japan Window is US users, we expect many Japanese to log on as well; US users who understand Japanese will also find the original sources useful in their research, studies or business. Although there are multi-lingual browsers available, the most inhibiting factor in placing Japanese text on-line is that often times the Japanese-language source is not available in electronic form. The staff of Japan Window is using Japanese OCR (optical character recognition) as much as possible to scan these important documents in, but the overhead involved in proofing the documents (as current OCR is about 85% accurate) makes it very time-consuming to place as much on-line as we would like.
6.0 Technical Approach
The conference presentation will discuss and preview some of the technical innovations which the Japan Window team is planning to debut. These include the following:
Major Technical Features
Multiple Entry Points. Since Japan Window will maintain large amounts of information updated routinely, it is important to provide users easy ways to access the information desired. In addition to the standard "subject area entryway", we will also utilize interactive maps and WAIS/database search so that a user may 1) quickly navigate his or her way through the Web site, and 2) go directly to the desired information items. For example, for someone trying to find the name and address of the Mayor of Osaka City, they could either click through the government pages, or traverse through detailed country, city, and area maps down to the city hall building. Another advantage in providing maps is to inform the user locations of important organizations. With an on-line "atlas", a businessman could identify the best way to get to a meeting, and a student could learn where to get important visa documents processed.
Custom User Interfaces. Since Japan Window is aiming at the high end user, with a high level of repeat usage, we are currently working on "customized" pages for particular users. Once the user has registered, Japan Window will provide pages and links which match best with that user's interests. For future "logons", Japan Window would recognize the domain name, provide a welcome message and indicate what's new on the system since the user last visited. This would increase the efficiency with which the user can obtain information. Of course, since SLIP connections produce temporary IP addresses and some domain hosts have multiple users, there will be a way for a user to "logon" with a particular ID and password, and get the resulting custom Japan Window pages.
Navigation Tracking. One of the biggest problems facing WWW providers offering extensive service or information is the inability to track which portions of a Web site have been visited by a particular user. As a result, a substantial amount of Web material that has been painstakingly prepared is simply not visited due to its inaccessibility or poor visibility.
We propose two mechanisms with which a Web provider can use to track a user's usage and navigation habits, hence allowing a very deliberate and controlled way of disseminating information that would otherwise be missed.
Mechanism A -- Passive Tracking. The first mechanism, being passive and indirect, is merely a creative use of existing access logs found in all known HTTP servers. Past sessions of a registered user provides important "personal footprints", i.e. information about which pages the user has taken interest in and which areas the user might have missed. Many existing log-analyzers (Kevin Hughes --- GetStat) have taken much pain out of deciphering and filtering the logs for the exact information that one needs. With this information, the Web provider would be able to customize his pages for the next time the same user visits the site. Thus, passive tracking can be implemented with minimal design overhead and is ideal for sites which carry extensive information and expects repeated usage.
Mechanism B -- Dynamic tracking. The second mechanism offers real-time tracking of a user's navigational progress. It allows the Web provider to dynamically alters his Web pages according to the needs of each user. Dynamic tracking is intended for use by interactive sites and sites that wishes to make immediate impact with their pages. The emphasis of such sites is not on a user's repeated usage of their pages but on a dramatic presentation of what is bringing out what is fresh and interesting.
As it would be impractical to track if each individual page of a site is accessed or not, a decision might have to be made regarding which pages should be tracked. Thus, a careful analysis of the Web structure of the site, possibly with the help of log data already gathered, will be needed to determine the "pivots" in the Web. There are many criteria to consider when determining the pivots. A few of which could be:
Some form of heuristics would probably have to be employed to determine the pivots. However, a page that satisfies the first criterion often would satisfy one or both of the rest too.
Implementation. In order to present information based on a user's past experience with the site, it is necessary for each page to take on multiple personalities. The scripting and server-side-include capability of http servers enables a page to contain dynamically generated hypertext references (hrefs). How the references are generated depends on access information which a site keeps of its registered users. We would suggest managing such information with a light-weight and highly efficient database. In fact, if a site does not expect a large number of users at one time, a judicial use of environmental variables can be used to track the access-state of a user in real-time.
Community-Serviceable Databases. Because Japan Window is expected to attract a large number of members who wish to share their activities with the larger community, management of new information can quickly become a challenge. To explore new methods for dynamic information management, Japan Window is currently evaluating the notion of community-serviceable databases. This approach is also being taken at Stanford's Mechanical Engineering Department for vendor information management [Hong et al, '94]. At present, Japan Window has implemented a prototype dynamic events calendar with such a model in mind.
The Japan Window Calendar. Our current prototype is implemented as a custom CGI-database using TclX. The interface model for the calendar is that of a filtered list which allows screening of events by month or sponsoring organization. To deal with the dynamic nature of event information, all calendar pages and input forms are "virtual" pages generated on the fly. In other words, the Japan Window server does not store any HTML files for calendar events. Instead, events and sponsor information are stored in database files from which data is extracted and reformatted to display information of interest. This approach provides greater flexibility in making changes to the events database, which in turn facilitates self-maintenance by the community.
Community Access Modes There are basically two modes of accessing the calendar: viewing and editing. Viewing of calendar events is accessible to the general public. Editing of calendar events, however, is restricted to Japan Window members, through password access. New members can sign up and set their passwords through an auxiliary forms interface. Once a member logs in, they are only allowed to edit events sponsored by themselves. In this manner, everyone can contribute to the database without accidentally jeopardizing another party's information.
Maintenance. In spite of our intentions to have the community maintain this database as a shared resource, problems are bound to happen, and a Web-based backdoor has been provided for carte blanch staff editing. Though similar maintenance could be done by a system administrator editing the raw data files, the Web provides an opportunity to build a friendlier interface, and such a "staff" interface has been implemented as a third mode of access for the Japan Window Calendar. This facilitates assignment of maintenance responsibilities to existing secretarial staff, most of whom are not UNIX savvy.
Web Site Administration: Japan Window "Cockpit". Japan Window is intended to be a highly automated Web site, with background software taking care of most daily site administration functions. The Japan Window "Cockpit" page, accessible only by Japan Window site administrators, provides a single point-of-entry and centralized control board for all operations such as database updates, status reports, question and comments answering, user accesses statistics, WAIS administration, etc.. We recommend that Web sites with complex information content adopt this approach, and automate as much as possible of their operations.
7.0 Conclusions and Future Work
Japan Window will debut simultaneously in the US and Japan in the very near future. Stay tuned to the NTT and Stanford Web sites, and the NCSA and CERN "What's New" lists for formal announcements. Organizations interested in collaborating in the Japan Window project should contact either the Stanford US-Japan Technology Management Center (blee@fuji.stanford.edu) or NTT (atsuhiro@nttam.com) for further details.
References
[1] Lee, B., Bayle, M. "Internet X-Guide to Japan Information Exchange," Technical Requirements of Accessing Japanese Information: Problems and Solutions, Library of Congress, Washington D.C., March 18, 1994.
[2] Hong, J., Toye, G. and Leifer, L., "Using the World-Wide Web for an Engineering Design-Team Class," Proceedings of the 2nd International WWW Conference, Chicago, IL, 1994.
Biographies
Mr. Burton Lee, Project Manager-US, is a Ph.D. candidate in the Department of Mechanical Engineering at Stanford University. He is principal designer and architect of the Stanford "X-Guide" Web site (http://fuji.stanford.edu), and maintains the "WWW Virtual Library: Japan" page. His research interests are in international manufacturing, networking, geographic information systems, and aerospace engineering.
Mr. Michael Bayle, Associate Project Manager-US, is an M.S. candidate in the Department of Computer Science at Stanford University. He is co-designer of the Stanford "X-Guide to Japan Information Resources" Web site, and is pursuing graduate research in human-computer interaction (HCI). His research interests include multi-lingual computing, electronic commerce, and virtual reality.
Dr. Atsuhiro Goto, Project Manager-Japan, is a Senior Research Engineer, Supervisor, with NTT Software Laboratories.
Mr. Yasuhiro Sakamoto, Associate Project Manager-Japan, is a software engineer with NTT Software Laboratories.
Mr. Toshihiro Takada is a researcer with NTT Basic Research Laboratories and author of the NTT multi-localization version of Mosaic. He presented a paper on multi-lingual Mosaic at the WWW Cern conference in May 1994, and is co-designer of the NTT Web site in Japan (http://www.ntt.jp).
Mr. Mitsukazu Washisaka is a research with NTT Basic Research Laboratories.
Mr. Jack Hong is a Ph.D. candidate in the Department of Mechanical Engineering at Stanford University. He is principal architect of the ME Department Center for Design Research Web site (http://cdr.stanford.edu), and maintains research interests in interactive design.
Mr. Vui Chiap Lam is an undergraduate in the Department of Computer Science at Stanford University. His interests include computer graphics and human-computer interaction.
Ms. Ayako Yamazaki is a staff member of the Japan Window Project. Her interests include linguistics and the teaching of Japanese to children.
Ms. Kathryn Kasa is an undergraduate at the University of California - Berkeley.
Contact Author: blee@fuji.stanford.edu