See Carol's pserv6.cht.memo (Mounted at /big on jasper) around line 6,737 for the start of her learning around March, 2002. See also /afs/d/project/search/tools/db2/fixfref2.pl for the program that fixes Japanese patent numbers in our fref load file. ---------------------------------------------------------------------------- See also http://training.dialog.com/quick/solutions/ss_pdfs/4300.pdf for Dialog's tutorial on Japanese patent numbers. Some tidbits from that are, * Searching for Japanese Patents is complex because the number you are searching could represent several different inventions. Inventions filed in Japan move through several levels of publication before eventually being published as a patent. With each publication, the Japanese Patent Office assigns a different number. Prior to 2000, this number was based on their Year of the Emperor and not the Western Year (Gregorian Calendar). Therefore, if you don't know exactly what type of number you have, you may have to search for the number several different ways before finding the correct invention. For example, in 1990 the number 2-63320 was a patent application number, a publication number of an unexamined patent application, and a publication number of an examined patent application. * The number you have can refer to any one of six Japanese patent documents types: 1. Patent Application 2. Published ("laid open") Unexamined Patent Application (Kokai) 3. Published Examined Application (Kokoku) 4. Granted Patent (Toroku) 5. Utility Model Application 6. Utility Model (Utility Model Numbers are NOT searchable on Dialog) ---------------------------------------------------------------------------- In our Patent database, we follow Japan's rules for patent numbers, using the Imperial year for pre-2000 patents and the Gregorian year for 2000 on (with the 4-digit year compressed like US Applications are, e.g. JP2004012345A1 = JP24012345A1). Two decent reference sites are http://www.i18nguy.com/l10n/emperor-date.html (that's i-eighteen-nguy.com/el-ten-en ... and http://www.globalsecurity.org/military/world/japan/year.htm Emperor Era Years Imperial Years To Convert Transition Date ========= ======= ========= ============== ========== =============== Keio 1866-1868 1-4 +- 1864 Mutsuhito Meiji 1868-1912 1-45 +- 1867 Yoshihito Taishou 1912-1926 1-15 +- 1911 1912-07-30 Hirohito Showa 1926-1989 1-64 +- 1925 1926-12-25 Akihito Heisei 1989-1999 1-11 +- 1988 1989-01-08 Actually, the Heisei era continues, and you could refer to the year 2000 as Heisei 12, but in 2000, the Japanese patent office started using the Gregorian year, so you won't see this in the Japanese patent data. We have Japanese patent data from 1959 on, which is convenient because there is no overlap of Imperial years. Years Imperial Years ========= ============== 1959-1989 34-64 1989-1999 01-11 Theoretically, there may be a case where an old patent refers to a really old Japanese patent, so that there could be ambiguity in the Imperial year. For example, does Imperial year 1 refer to the Showa era (so, 1926) or the Heisei era (so, 1989)? You need to know the era. On the patent image, the Imperial Year will be separated from the number by a dash, e.g. 6-123456. Another ambiguity might be when you're given the number without the dash, e.g. 6123456. Is this 6-123456 from 1994, or 61-23456 from 1986? ---------------------------------------------------------------------------- The raid.espace table in the Thomson Patent Store (TPS) world, uses this format for pre-2000 Japanese patents, EYYNNNNNN, where E = The Era (2=Taishou, 3=Hirohito, 4=Heisei) YY = Imperial Year within that Era and NNNNNN = the number Delphion's # raid.espace:pn Example Year = Delphion # = TPS # Format ============ =============== ============================================ JP34 - JP64 Add 300,000,000 48-10930 (1971) = JP48010930 = JP_348010930 JP01 - JP11 Add 400,000,000 11-515157 (1999) = JP11515157 = JP_411515157 For post-2000 Japanese patents, the format is YYYYNNNNN, but there are a lot of apparent duplicates, for example, CC PN DT URL -- ---------- -- ----------------- JP 2000000001 A JP2000000001A J JP 2000000001 A JP2000000001A AJ JP 2000000001 U JP2000000001U J I forgot what Joachim told me, but TPS ignores some of these rows. The Japanese kinds one finds in raid.espace are all over the map and I don't understand. Kind Count ==== ========== A 14,210,009 B 726,797 U 483,037 B2 2,659 Y2 1,442 T1 1,029 Y1 848 B1 708 And 17 others kinds all with count<25. ---------------------------------------------------------------------------- On 2-10-2005, TPS's raid.done table had 7,071,655 JP rows, all with kind=A1. Whereas the Southbury imag table had 6,786,529 JP rows, all with kind=A2. The raid.done table stores pre-2000 Japanese rows with the same number as Delphion (yynnnnnn) and post-2000 rows with yyyynnnnnn. TPS pn in TPS Count SBY Count SBY imag patn Year raid.done raid.done imag (if different) ==== ========== ========= ========= ============== 1976 51nnnnnn 24,800 24,460 1977 52nnnnnn 77,869 76,783 1978 53nnnnnn 70,202 69,323 1979 54nnnnnn 80,096 79,071 1980 55nnnnnn 104,898 103,334 1981 56nnnnnn 109,425 107,622 1982 57nnnnnn 139,267 137,029 1983 58nnnnnn 151,789 149,673 1984 59nnnnnn 157,763 155,133 1985 60nnnnnn 182,716 180,545 1986 61nnnnnn 207,684 205,273 1987 62nnnnnn 205,795 199,102 1988 63nnnnnn 224,638 218,153 1989 64nnnnnn 3,018 2,571 ???? 11,102 00nnnnnn <--- What are these??? 1989 1nnnnnn 309,315 260,646 1990 2nnnnnn 296,695 262,443 1991 3nnnnnn 290,141 267,993 1992 4nnnnnn 366,879 340,944 1993 5nnnnnn 341,623 315,260 1994 6nnnnnn 344,280 319,885 1995 7nnnnnn 332,033 310,422 1996 8nnnnnn 336,213 317,148 1997 9nnnnnn 302,388 260,486 1998 10nnnnnn 341,388 341,388 1999 11nnnnnn 355,942 355,942 2000 2000nnnnnn 358,084 358,084 20nnnnnn 2001 2001nnnnnn 359,103 359,103 21nnnnnn 2002 2002nnnnnn 374,551 374,551 22nnnnnn 2003 2003nnnnnn 348,061 348,061 23nnnnnn 2004 2004nnnnnn 274,999 274,999 24nnnnnn ----------------------------------------------------------------------------