Ex-post evaluation by bibliometric method. Institutional comparison among 9 resembled foreign research institutes by using the energy citation database (ECD)

From a viewpoint of ex-post evaluation, research papers published from nine resembled nuclear research institutes located in Japan, the U.S.A., Germany and France were compared by a bibliometric method. A research database used was the Energy Citation Database (ECD) owned by USDOE. ECD is a database run by USDOE and has a high frequency of research paper acquisition assembled in the U.S. Response speed of ECD on the Website is quick and all logged data can be handled easily. INIS database is run by the International Atomic Energy Agency (IAEA) and contains a lot of nuclear research papers collected from the member countries such as the U.S.A., Japan, Germany and France. INIS underestimates about 20% of the U.S.A. data than that of ECD. I. Institutional Comparison. (1) ECD shows that a total number of research papers published during 25 years (1978-2002) was of the order of the ORNL (34, 149 papers)>SNL>ANL>BNL>Idaho (>Karlsruhe>JAERI>Jeulich>Cadarache). Where, INIS shows it as ORNL>JAERI. (2) ECD can show a long-term data comparison with a time span more than 50 years (1953-2002). Disclosed research papers were of the order of the ORNL (55,857)>ANL (37,129)>SNL (24,628)>BNL (24,829)> Idaho (2,398). There were many records loaded without publication dates-over 50,000. Because of this, any searches which use dates are not finding these documents. Typically, the author found over 5,000 SNL items in the NSA range of records. SNL also kept a lot of defense reports, those are not disclosed yet. One had better know a historical background of each cite as to the case for long-range dates comparison. (3) ECD founds that research papers at a five-year period varied those numbers. At past (10), thus 1988-1922, paper reduction occurred sharply at most US-institutes. This might be attributed to lay-offs, funding shifts or complete elimination of programs, a policy change in reporting requirements for contract reporting deliverables. Definitions of what constituted STI (science and technology information) were beginning to vary also. II. Comparison at the research field of material science. With time span of 25 years, a comparison was made in the U.S. (1) Actinides; a number of papers disclosed was of the order of ORNL (944 papers) > ANL (927) > SNL (255). The difference is negligible between ORNL and ANL. ORNL was also the champion for the case of INIS. (2) Electromagnetic radiation; this case was of the order of ORNL (1.043 papers) > BNL (673) > SNL (605). ORNL was a champion, too. The activity of radiation application represented by this keyword might be degraded in the U.S. around a year of 1993. For INIS comparison, it was of the order of JAERI > ORNL > BNL. The champion was JAERI. (3) Neutron; it was of the order of ORNL (3,934 papers) > ANL (2718) > BNL (1568). The champion was ORNL > For the case of INIS, it was JAERI > ORNL > BNL. The neutron study in JAERI was most advanced. III. Comparison by CPI. CPI shows that ORNL research activities were beginning to vary from time to time. Covering period from past (20) to past (10), ORNL activity indicated by CPI was high but was low at the present (1998-2002). The tendency coincided with that observed in INIS. (author)