Introduction

The host computer of the National Diet Library (NDL) is close to an end.  When it hands over its task to the new Digital Library Infrastructure System in January 2003, I would like to reward it by thinking back over its service for thirty years from a system developer's eyes.

1. Childhood - the early days of Japanese processing

"State-of-the-art computer equipment is crammed into in a small room on the fifth floor in the NDL.  Now, Librarian Mr. Kubota pushes the start button..."

It was on December 19, 1970.  Our first machines were a general-purpose computer (HITAC 8400) by Hitachi, Ltd., kanji printers, and a kanji entry teletype machine.  The main memory of the CPU was 65 KB, 1/10,000 the capacity of a PC terminal in our new system. The external magnetic disk storage capacity was 29 MB, 1/1,400 of the internal hard disk drive capacity (40 GB) in a PC terminal used in the new system.  In terms of size, the new PC occupies just a part of a desk whereas the old machine was several hundred kilos of metal that came in a 4 ton truck. 

Around 1972, Japan was in the final stage of the high economic growth period.  Public and private offices were scrambling to introduce computers.  In the IBM-led computer market, six domestic firms engaged in fierce competition.  As soon as a computer firm succeeded in a bid, it rushed manufacturing around the clock toward the delivery date.  At the same time, a client also pressed on with program creation to activate the machine on arrival. 

At the NDL, our staff members too developed an initial system for the "General Index for the Minutes of the Diet" in less than a year with system engineers of the computer manufacturer.  It's a matter of course now, but kanji processing was totally undeveloped at that time.   We faced numerous problems in processing Japanese containing kanji and solved them one by one with the manufacturer.  That was a real pioneer work. 

The Librarian attending the opening ceremony of the NDL computer room (1970)

2. Adolescence - The age of catalog and index compilation 

In the 1970s, we devoted all the time to developing systems one after another such as catalog editing and index editing based on the already acquired know-how of kanji processing.  Above all, the system for Japanese books, which was built from 1976 to 1978, handled the main collection of the NDL and subsequently became the most important system in hatching the JAPAN/MARC. 

Newly employed or transferred staff were pushed into technological training at once, returned to develop a program, then went into another training program.  As we had constant challenges to face, some people did not like to be assigned to our division.  However, fortunately many young people were in the office so we could have a lot of fun as well. 

The kanji printer was one of the leading features of the age.  We actually introduced two different kanji printers.  When we first introduced the machines, the total cost of rental was eight million yen per month, and nearly 40 percent of it was spent on the kanji printers.  They carried an up-to-date character generator, but still they required a lot of care.  For example, one printer needed special printing paper that was really troublesome.  The paper changed color in time and rolled up when it was cut and dried out.  When we tried piling up the sheets and weighing them down before they were dried, they stuck to each other.  If we tried to peel them off, they would be partly torn.   However, the printed kanji were so beautiful that we were often impressed by the high quality of printing.

Kanji printer

3. The prime of life - The age of JAPAN/MARC 

The host computer was most active in the 1980s.  The symbolic results in this age were JAPAN/MARC, NOREN and J-BISC.  In the late 1970s, many libraries were promoting computerization and they began to be quite eager for bibliographic information as a basic resource.  The NDL JAPAN/MARC council started in 1979.  It reported on format specifications and usage of JAPAN/MARC in the following year.  Extreme arguments flew about in the discussion and we eventually decided to produce JAPAN/MARC from the system for Japanese books and made it conform to UNIMARC as much as possible.  In this way, the long-hoped-for JAPAN/MARC (Books) distribution started in April 1981.  Soon after that, it was used in Canada, the U.S. and Australia and was appraised internationally. 

Almost at the same time as the launch of the NDL JAPAN/MARC council, we began to consider introducing an online information retrieval system.  Knowing that Hitachi would release "ORION", an online retrieval system with a Japanese environment, the NDL decided to adopt it.  "ORION" was produced with extended "BASIS", which was an online information retrieval system developed by Battele (U.S.) that was spreading all over the U.S. in those days. "ORION" was formally introduced in autumn in 1980.  It made remote search possible and attracted the attentions of public libraries throughout Japan. 

"ORION" was a brand name, so we invited the NDL people to come up with a nickname.  In the end, we named it "NOREN" (National Diet Library Online Information Retrieval Network System).  Most of the systems in Japan had Western style names at that time so we really wanted a Japanese name which sounded traditional.  As a result, it seems a bit far-fetched.  (Translator's note: Noren is a short split curtain hanging over the doorway of Japanese traditional shops such as Sushi bars and Soba noodle restaurants to show that they are open for business.)  Later on the NOREN database pattern was applied to any information handled by the NDL and its variations got nearly forty. 

JAPAN/MARC distributes bibliographic information in magnetic tape form and people have to install some devices to use it.  That's why JAPAN/MARC users were limited to university libraries, large public libraries and information service institutions. 
In contrast, J-BISC in a CD-ROM form is available with a personal computer, so sales were rising rapidly since it was released in 1988.  The number of users was close to 1,200 at the peak time in 1997, among which nearly 100 were elementary or junior high school libraries.

Along with the J-BISC, the CD-ROM Line series was released in July 1994.  The series consists of the National Union Catalog of Braille & Recorded Books in Japan, the National Diet Library Catalog (Meiji, Taisho Period and 1926-1945), the Japanese Periodicals Index and the National Diet Library Catalog of Serials. 

NOREN and the Input and Correcting Management System (ICMS) were the two main systems established in the 1980s.  ICMS was a data input and proof system for external punchers inputting and correcting data intensively.  They deal with an average of twenty million characters per year so we should not forget their significant contribution to the basic databases of the NDL.  The NDL staff also created various kinds of products from the input data, and the value of the products was maximized when they were carried on the NOREN.  As NOREN has many features and it could be easily linked with a user program, we used it like a database management system. 

Punchers' office

Kanji keyboard