Site: National Center for Science Information Systems
3-29-1 Otsuka, Bunkyo-Ku
Tokyo 112, Japan
Date visited: May 26, 1995
Report Author: E. P. Glinert
E. P. Glinert
The National Center for Science Information Systems (NACSIS) is a unit of the Ministry of Education, Science, and Culture. It was formed in April 1986 from the Center for Bibliographic Information at the University of Tokyo and is located on the old campus of Tsukuba University. It is one of about 15 interuniversity research institutes that serve all university researchers in Japan. Its charter is to gather, organize and provide scholarly information, as well as to carry out R&D related to scholarly information systems. Its scope spans the natural sciences, the social sciences, and the humanities. Its network services link university libraries, computer centers, information processing centers, and national university research institutions throughout Japan, and interconnect them with research networks in the United States via a node in Stockton, California and in the United Kingdom via a node at the British Library in London.
By the end of fiscal year 1993, a total of 310 university libraries were connected on-line to NACSIS at 243 institutions. There were 44 databases in service containing over 44 million records. The primary NACSIS-CAT unified holdings catalog alone contained about 12.9 million records. The NACSIS-ILL inter-library loan service recorded more than 400,000 transactions among the 190 participating institutions during the year. NACSIS inaugurated the Japanese national electronic mail service in 1987, which it continues to support and enhance (e.g., network news was added in Fiscal Year 1993). Extensive research and development are carried out as part of the organization's activities; current topics of interest are described below, and include digital libraries and high-performance communication networks for scientific research. Education and training (e.g., on network operation, database construction, and cataloging) are also provided.
Dr. Hisao Yamada's interests focus on cognitive science aspects of specialized task performance. He has long been concerned by Japan's low office productivity due to problems associated with text processing. Typically, the user must type in katakana (phonetic) symbols; software then generates the appropriate kanji. But office workers must read kanji as they type katakana, and often they don't know how to properly pronounce the former because they have not learned to read or write well! Furthermore, there are often multiple words that sound the same, so the software will have difficulty finding the correct kanji for the given katakana.
Dr. Yamada's approach to remedying these problems is based on the observation that vertebrate brains have been processing patterns for about a thousand times as long as they have been processing language. Converting information consciously from pattern to language can be difficult. Why not skip that step by viewing the typing activity as transferring patterns to paper (or whatever medium)? He devised a "two-stroke code" to teach typing without involving any mnemonic/language associations, which in tests achieved good results among young female subjects, but which required about 200 hours of training and so is not widely used, despite unpublished work by Ricoh that establishes the method's superiority over the standard multishift system.
With regard to the great deal of work currently being done throughout Japan on voice input, in the hope of circumventing the problems associated with keyboards, Dr. Yamada noted that the problems that make it hard for software to automatically translate from katakana to kanji also make voice input a much more challenging task than some researchers would like to admit. Yet another problem here is that the human voice changes during extended work sessions, as the speaker becomes tired or bored.
Dr. Yamada gave the JTEC team a brochure that provides a more global overview of the research sponsored by NACSIS and enumerates specific recent projects with PI names but without other details such as funding level or abstracts of publications. The research seems to fall into four broad categories: information science research (including standardization of bibliographic data and cataloging, automation of library housekeeping functions, automatic keyword extraction from texts, concurrent support for multiple East Asian character sets, multimedia databases, etc.); systems research (including hardware and software for large-scale databases, authority control, integrated management of primary and secondary information, electronic mail, fax transmission, CD-ROM publishing, usable human interface development, architectures and mechanisms for optimal acquisition of foreign scholarly information and for distribution of domestic scholarly accomplishments overseas, etc.); research trends; and project-type research activities. Examples of the latter are research on a digital archive for music, study of science information flow through networks, and development of a high-performance communication network for scientific research.
During the discussion, Dr. Yamada also addressed issues related to research in general in Japan, which he feels is harder to carry out, especially in universities, compared to the United States. He sees use in Japan of the Internet growing in the future, but lagging behind that of the United States for some time, largely due to the "lousy" writing system that leads to keyboard avoidance. Transfer of old materials to the new media will be difficult or even impossible, because many government documents are discarded after several years. Indeed, government officials are, on average, rotated out every 3 years as well, so there is a lack of expertise for decisions to make long-term changes (although the writing system has been simplified, or reformed, several times during the recent past). Companies adopt a slightly longer view, because people stay with the same organization for life, but even here research managers' positions last only 5 years or less and then rotate, so there is little planning beyond that range.
NACSIS. 1995. National Center for Science Information Systems 1994/95. Brochure. Okadome, T., and H. Yamada. 1990. A comparative study of input methods for Japanese text typing. Computer Processing of Chinese and Oriental Languages 4(4)(March):275-294.
Tatuoka, H. and H. Yamada. 1994. An input system for Vietnamese language text. Proceedings of the International Conference on Computer Processing of Oriental Languages, May 10-13, Taejon, Korea.
Tatuoka, H., and H. Yamada. 1995. Toward the Japanese input through Roman orthography. Submitted to ICCPOL '95, 23-25 November, Honolulu.
Yamada, H. 1980. A historical study of typewriters and typing methods, from the position of planning Japanese parallels. Journal of Information Processing 2(4):175-202.