PDFファイル - kaigi.org

The 26th Annual Conference of the Japanese Society for Artificial Intelligence, 2012
4E1-OS-15-4
᪂Ϊঠ஦΅⁄‫⁉•⁜⁆‷‫‬ⁱ‫‮‬Ύ ​࿇ᮇᕷሙືྥ΅ศᯒ
Analysis of Long-term Market Trend by Text-Mining of News Articles
‫܍‬ᮏ ଇஂ*1
࿴Ἠ ₩*1
Takahisa Kuramoto
Kiyoshi Izumi
୰ᔱ ၨᾈ*2
Akihiro Nakashima
ྜྷᮧ ᚸ*1
Shinobu Yoshimura
ᯇ஭ ‫ܘ‬஬൫*3
Tohgoroh Matsui
*1ូ
▼⏣ ᬛஓ*2
Tomonari Ishida
ྜྷ⏣ ⛱*4
Minoru Yoshida
୰ᕝ ࣡ᚿ*4
Hiroshi Nakagawa
ᮾி኱Ꮫ኱Ꮫ၄ូ ᕤᏛDŽ◊"⛉
School of Engineering, The University of Tokyo
*2
ญᮧ੷ๆᰴᘧ఍♫ᬌ
Nomura Securities Co., Ltd
*3
୰ൺ኱Ꮫូ ⏕࿨೺ᗣ⛉Ꮫൺ
College of Life and Health Sciences , Chubu University
*4
ᮾி኱Ꮫូ ᝟ሗᇶ┙‹ⁱ‽⁺
Information Technology Center, The University of Tokyo
Abstract: In this study, we developed a new method of the long-term market analysis by using text-mining of news
articles. Using our method, we conducted extrapolation tests to predict stock price averages by 19 industry and two market
averages, TOPIX and Nikkei225 for about 10 years. As a result, 8 sectors in 21 sectors (about 40%) showed over about 60%
accuracy, and 15 sectors in 21 sectors (over 70%) showed over about 55% accuracy. We also developed a web system of
financial text-mining based on our method for financial professionals.
1. `Ῐ Ύ
άືΎᙺWῤ᝟ሗ‒ᥦ౪Ῑ​⃺
Ȃ῭΅ኚື`Ȃ῭ᣦᶆ ᰴ౯Ύ௦ࢢ῕‌​ᩘ್⁅⁺‽⃸⁉⁣⁺
‷ ᪂ΪΎ௦ࢢ῕‌​⁄‫⁅⁆‷‫‬⁺‽ῪῩ⃸Ѯ኱Ὺ᝟ሗῌ┦஫Ύస
⏝ῗྜΰῦỴᐃ῕‌​῟ ჰɪᙧῪᣲື‒♧Ῑ⃺῝‌Ύᑐῗῦᢞ
ଘᐙ`⁄‭⁉
⁩ศᯒ ⁓‟ⁱ‾ ⁱ‽⁩‸ศᯒ΅ ῆῪศᯒᡭἲ
‒ቁ౑ῗῦᢞଘ‒ࢌΰῦῄ​ῌ⃸῝‌ῧ ࿇ᮇ΅ኚື΅ண `ჰ
ᖖΎᅔႤῪ΅ῌ⌧≧ῧῂ​⃺ΐ‌`᝟ሗฎ΅Ѯ኱῕⃸᝟ሗ࿛΅࿥
ಀᛶ΅ࣽ႗῕ΎୣᅉῗῦῊ ⃸῝΅᝟ሗ඲ῦ‒▐᫬Ύॾจῗῦപ
ษῪᢞଘ‒ࢌῆΐῨ`୙ྍЏΎಶῄ⃺࿇ᮇ΅ኚືΎῊῄῦ`⃸ኚື
‒ỴᐃῙ​᝟ሗ ῕ ΎѮ኱ΎῪ ⃸῝΅࿥ಀᛶ ୙᫂░ΎῪ​῟
⃸ᢞଘุ᩿`┈ᾫႤῗ῏Ὺ​⃺
ΐ‌῾ῧ΅◊"ῧ`⁉⁣⁺⁧⁩⁋⁁⁆⁺‭>዇ @ ഷఏ
ⓗ⁕‫⁝⁧‮‬ⁱ‫>‮‬ఀᗞ @Ῠῄΰ῟ᶵᲔᏛ͖΅ᡭἲΎ ΰῦȂ
῭΅ኚື‒ҜືⓗΎศᯒῙ​ΐῨ‒┠ᣦῗ῟ ΅ῌᏑᅾῙ​⃺ῗΉ
ῗ⃸ᩘ್᝟ሗ΅ॾᯒȆᯝ΅῿ῧ࿇ᮇⓗῪᢞଘุ᩿‒ୗῙΐῨ`Ⴄ
ῗῄ⃺῝΅⌮⏤Ῠῗῦ⃸ே࿛ῌॾจ‒ࢌῈῪῄΏῩࣽ႗Ὺ᭱പ໬ῌ
⏕Ῐ​ΐῨ ⃸ᩘ್እ΅᝟ሗῌḞؐῗῦῗ῾ΰῦῄ​ΐῨῌᣲῒ ‌
​⃺῝ΐῧ⁄‫⁆‷‫‬᝟ሗ‒⏝ῄῦȂ῭΅ኚື‒ศᯒῙ​◊"ῌಶᖺ
⌧‌ῦ῍῟⃺࿴Ἠ ` &35 ἲῨῄῆᡭἲ‒࿔Ⓨῗ⃸⁄‫ॾ‒⁆‷‫‬ᯒ
Ῑ​ΐῨῧȂ῭ືྥ΅ศᯒ‒ࢌΰῦῄ​>࿴Ἠ @⃺⁄‫⏝‒⁆‷‫‬
ῄ​฼ⅬῨῗῦ`ॾจῌᐜ᫆ῧῂ​ΐῨ⃸ᩘ್Ύྵ῾‌Ὺῄ᝟ሗ ศᯒῧ῍​ΐῨ΅ Ⅼῌᣲῒ ‌​⃺
ᮏ◊"ῧ`⁄‫⁆‷‫‬᝟ሗ‒⏝ῄ⃸ᢞଘᐙΉ ΅ोᮃῌ዇ῄ࿇ᮇ
ⓗῪȂ῭ືྥ΅ศᯒ‒ࢌῆ⃺῕ ΎѮ኱Ὺ᝟ሗ΅୰Ή ᢞଘΎ᭷
┈Ὺ᝟ሗ‒ᢳฟῗ⃸῝΅᝟ሗ࿛΅࿥ೳᛶ‒᫂☜ΎῙ​ΐῨῧᢞଘ
2. ศᯒᡭἲ
ᬌᮏ◊"΅ෆᐜ`ญᮧ੷ๆᰴᘧ఍♫΅බᘧ॒ॾ‒♧Ῑ ΅ῧ`
ῂ ῾Ί–⃺ូ
ೳȑඛ៖ូ ࿴Ἠ₩⃸ᮾி኱Ꮫ኱Ꮫ၄ᕤᏛDŽ◊"⛉‵‷⁄⁞๰ᡂᏛᑓ
ᨷ⃸Ᾰ៍៍៏៉។្៑្ូ ᩥி༊ᮏඃ ៓៉៏៉៍⃸ូ ᠅᠖᠑᠉᠅ៜ᠏᠕᠏៊᠐៊᠑៉᠐᠋᠇᠕᠋៊៽៿៊᠆᠌ូ
ᮏ◊"ῧ`࿴Ἠ ῌᥦၐῙ​ CPR ἲ‒⏝ῄῦศᯒ‒ࢌῆ⃺CPR
ἲῨ`ඹୣॾᯒ⃸୺ᡂศศᯒ⃸ᅇᖐศᯒ΅୕ẁၥΉ Ὺ​ศᯒᡭ
ἲῧῂ​[࿴Ἠ 2011]⃺ᚑ᮶΅ CPR ἲῧධຊ⁄‫⁆‷‫‬Ῠῗῦῄ῟᪥
ᮏຏࢌ΅ฐࠕȂ῭᭶ሗ`ẚుⓗᙧᘧ΅Ỵ῾ΰ῟ᩥ᭩ῧῂ​῟ ⃸
ჰᖖΎᢅῄ Ῑῄ ΅ῧῂΰ῟⃺ᮏ◊"ῧ`᪂Ϊঠ஦Ῠῄῆᙧᘧ
ῌฐࠕȂ῭᭶ሗ ᐃ῾ΰῦῄῪῄᩥf‒⏝ῄ⃸ ᗈúῧ࿇ᮇ
ⓗῪศᯒῌྍЏῨῪ​ ῆΎ CPR ἲ‒ᣑᙇῗ῟⃺
2.1 ඹୣॾᯒ
ᮏ◊"ῧ⏝ῄ῟⁄‫⁆‷‫‬᝟ሗ`᪥ᮏȂ῭᪂Ϊῧῂ​⃺Ȃ῭ǜῧ
ῂ​῟ ⃸Ȃ῭΅ኚື‒ỴᐃῙ​ोᅉῌᥖ౅῕‌ῦῄ​ῨͳῈ ‌
​⃺
῾Ὶᆅ᪉ჳ‒၆῏඲ঠ஦Ύᑐῗῦ ChaSen[ChaS]‒⏝ῄῦᙧែǢ
ॾᯒ‒ࢌῄ⃸ື৒⁹ྡ৒⁹ᙧᐜ৒‒ᢳฟῙ​⃺῝ῗῦྠ୍ᩥ୰Ύၴ
᥋ῗῦฟ⌧ῗ῟༢ਂ΅ῆῡ⃸ᑡῪ῏Ῠ ୍᪉Ύ᪥Ȃ‵※⁺⁧‷>ូ ᪥
[email protected]΅῿‒ᩘῈ
ୖῒ​⃺᪥Ȃ‵※⁺⁧‷Ῠ`᪥ᮏȂ῭⁅‶‽⁩ ⁅‡†ῌබ࿔ῗῦ
ῄ​᪂Ϊঠ஦᳨Ǥ΅῟ ΅಑᭩ῧῂ ⃸lj1 ୓3 ༓ਂῌ཰༃῕‌ῦ
ῄ​⃺༢ਂ῔Ῠ΅ฟ⌧ᆆᗘ‒ᩘῈୖῒ​ ⃸ၴ᥋ῗ῟ඹୣ࿥ಀ
΅ฟ⌧ᆆᗘ‒ᩘῈୖῒ​ΐῨῧȂ῭ືྥΎ࿥Ῑ​᝟ሗ‒ῆ῾῏ᢳฟ
ῧ ῍⃸ॾจ ᐜ᫆Ύ Ὺ ​ῨͳῈ ‌ ​⃺ᚑ᮶΅ CPR ἲῧ`
KeyGraph †⁩′⁨‸⁞[኱⃝ 2006]‒⏝ῄῦฌोਂ‒ồ ῦῄ​ῌ⃸
ᮏ◊"`᪂Ϊঠ஦‒⏝ῄ῟࿇ᮇண ῧῂ​῟ ⃸᪥Ȃ‵※⁺⁧‷
΅ਂῨ΅ඹୣ‒Ῠ​ΐῨῧɉ̦ⓗῪ᝟ሗ‒ᢳฟῗ῟⃺ΐ΅ᩘῈୖῒ
‒ ⁴᭶࿛΅ঠ஦ῧˎ ಹῗࢌῄ⃸࿸್௨ୖῪ ฟ⌧⃸࿸್ᮍ‶Ὺ
୙ฟ⌧Ῠῗ⃸ฟ⌧⁏‽⁺ⁱ‒ᐃ̿ῗ῟⃺ῪῊ⃸࿸್` Ῠῗ῟⃺1
⁴᭶࿛Ύฟ⌧ῗ῟ඹୣ࿥ಀ΅ῆῡᑡῪ῏Ῠ ୍᪉Ύ᪥Ȃ‵※⁺⁧‷
-1-
The 26th Annual Conference of the Japanese Society for Artificial Intelligence, 2012
2.2 ୺ᡂศศᯒ
2.1 ÷΅ඹୣॾᯒ‒ഓཤ 3 ᖺ࿛⃴36 ⁴᭶⃵΅᪂Ϊঠ஦Ύᑐῗῦࢌ
ῄ⃸࿸್௨ୖῪ ῰ 1⃸࿸್ᮍ‶Ὺ ῰ 0 Ῠῗ⃸1 ⁴᭶῔Ῠ΅༢ਂ΅
ฟ⌧⁏‽⁺ⁱ‒Ȇྜῗ῟ࢌิ‒సᡂῙ​⃺ΐ΅᫬⃸ছɾᮇ࿛ῧᑡῪ
῏Ῠ ୍ᅇ`ฟ⌧ῗ῟༢ਂᩘ`lj 2 ༓ਂῧῂ ⃸36 ࢌ 2000 ิ΅ࢌ
ิῌసᡂ῕‌῟ΐῨΎῪ​⃺ΐ΅ࢌิΎᑐῗῦ୺ᡂศศᯒ‒ࢌῄ⃸
ྛ᭶ 15 ಶ΅୺ᡂศῧঠ஦‒ো౯ῗ῟⃺ῙῪ‫‏‬ῡ⃸1 ⁴᭶࿛΅᪂Ϊ
ঠ஦‒ 15 ḟඖ΅⁗‭⁆⁩ῧো౯ῗ⃸῝΅⁗‭⁆⁩‒ȆྜῙ​ΐῨῧ
᪂Ϊঠ஦΅≉ᚩฎ΅᫬DŽิ⁅⁺‽ῌᚓ ‌῟ΐῨΎῪ​⃺
2.3 ᅇᖐศᯒ
ᰴ౯⁅⁺‽` 12085$ ΅ ᴗ✀⃸῕ Ύ᪥Ȃᖹᆒ⃸
723,; ‒⏝ῄ῟⃺NOMURA400 Ῠ`⃸ญᮧ੷ๆฐࠕᕤᏛ◊"‹
ⁱ‽⁺ῌᥦ౪ῗῦῄ​⁅⁺‽ῧῂ ⃸᪥ᮏᰴᘧᕷሙ΅඲ຝ᯶Ή വ
ᐃ῕‌῟ᕷሙ௦ࢢᛶ΅዇ῄᢞଘ⁤⁉⁎⁺‷ῧῂ​⃺ᵓᡂຝ᯶`᪥
ᮏᰴᘧᕷሙ΅඲ຝ᯶΅୰Ή ⃸†⁈⁨‷⁆΅ព॒‒ᇶΎവᐃ῕‌῟
ᕷሙ௦ࢢᛶ΅዇ῄ 400 ຝ᯶ῧῂ​[ญᮧ]⃺῝΅ῆῡ⃸໬Ꮫ⃸๢໠⁹ჰ
๢⃸ᶵᲔ⃸Ҝືత⃸Ⴑᶵ⁹Ɲᐦ⃸་⒪⁹⁖⁩‷ †⃸ᇇရ⃸ᐙᗞ⏝ရ⃸
ၟ♫⃸ᑠ኎ ⃸″⁺⁑‷⃸※⁓⁆․‥†⃸ ⁅‡†⃸೫ಙ⃸೫ಙᘓয⃸ఫ
Ꮿ⁹୙ື⏘⃸ഐ౪⃸බ┈⃸ฐࠕ΅ 19 ᴗ✀ῌᑐ૏ῧῂ​⃺᪥ᮏ΅ᕷሙ
΅ศᯒΎ`പῗῦῄ​ᣦᶆῠῨุ᩿ῗ῟⃺῾῟⃸ᕷሙ඲య΅ື῍‒
ᢕᥱῙ​῟ ΅ᣦᶆῨῗῦ᪥Ȃᖹᆒ⃸TOPIX ண ᣦᶆΎຍῈ῟⃺
ᣦᶆ݅ ΅᫬้‫ ݐ‬ΎῊῑ​ᰴ౯‒‫݅݌‬ǡ ‫ ݐ‬ῨῙ​Ῠ⃸༢఩ᮇ࿛᳃‫( ݐ‬1 ⁴᭶⃸2
⁴᭶⃸3 ⁴᭶)ῧ΅⁨‽⁺ⁱ‫݅ݎ‬ǡ ‫` ݐ‬ୗᘧῧᐃ̿ῧ῍​⃺
‫ݎ‬௜ǡ௧ ൌ
1 ⁴᭶ᚋ΅ண ΎῊῄῦ` 9 ᖺ 9 ⁴᭶΅ 117 ᅇ΅᥎ᐃ‒ࢌΰ῟⃺῾
῟⃸2 ⁴᭶ᚋ⁹3 ⁴᭶ᚋ΅ண ΅ሙྜ⃸⁨‽⁺ⁱΎῊῑ​༢఩ᮇ࿛᳃‫ݐ‬
‒ਚᩚῙ​ΐῨῧ᥎ᐃ‒ࢌΰ῟⃺
3.1 1 ⁴᭶ᚋ΅እᤄண Ȇᯝ
ូ እᤄண ΅ኀؐṇµ⋡΅Ȇᯝ‒ᅗ 1 Ύ♧Ῑ⃺ΐ΅Ȇᯝ`እᤄண
‒ࢌΰ῟ᅇᩘ΅ῆῡ⃸ኀؐῌ୍Ҥῗῦῄ῟ᅇᩘ΅๭ྜ‒ⓒศ⋡
ῧࢢῗ῟ ΅ῧῂ​⃺
1 ⁴᭶ᚋ΅ண ΅ኀؐṇµ⋡`ᕷሙ඲య΅ືྥ‒ࢢῙ TOPIX
᪥ȂᖹᆒῨῄΰ῟ᕷሙᖹᆒᰴ౯ῧ 63.7%ῧῂΰ῟⃺῾῟⃸ᢞଘุ
᩿Ὸ΅പ⏝ྍЏᛶ΅┠ᏳῨῪ​ 55%௨ୖ΅ṇµ⋡` 7 ๭௨ୖ΅ᴗ
✀(ண ᣦᶆ 21 ΅ῆῡ⃸15 ᣦᶆ)ῧങᡂῙ​ΐῨῌῧ῍῟⃺55%Ῠῄῆ
ኀؐṇµ⋡` AI ‒⏝ῄ῟ᐇၯ΅⁓‟ⁱ⁇ ┠ᣦῗῦῄ​ᩘ್ῧῂ
​[AERA]⃺ᮏ◊"ῧ`᪥ḟ ࿇ᮇ΅᭶ḟண ῧ⃸ῗΉ lj 10
ᖺ࿛Ῠῄῆ࿇ᮇ࿛΅እᤄண ⁄‷⁆ῧᏳᐃῗ῟Ȇᯝ‒♧ῗῦῄ​⃺ΐ
‌`ᢞଘᐙΉ ोᮃ΅዇ῄ࿇ᮇண ΅Ɲᗘ‒኱ᖜΎᨵၿῗ῟Ῠῄ
ῆΐῨῧῂ​⃺
ᅗ 1℆ូ 1 ⁴᭶ᚋ΅እᤄண ΎῊῑ​ኀؐṇµ⋡
ϲϰ͘ϭ
ϲϯ͘Ϯ
ϲϯ͘Ϯ
ϱϵ͘ϴ
ϱϵ͘ϴ ϱϵ͘ϴ
ϱϵ
ϲϬ͘ϳ
ϱϳ͘ϯ
ϱϳ͘ϯ
ϲϬ
ϱϱ͘ϲ
ϱϲ͘ϰ
ϱϲ͘ϰ
ϱϳ͘ϯ
ኀ
ϱϰ͘ϳ
ϱϱ͘ϲ
ؐ ϱϱ
ϱϮ͘ϭ
ṇ
ϰϵ͘ϲ
ϰϴ͘ϳ
ϰϳ͘ϵ
µ ϱϬ
ϲϱ
⋡
⃲
‫݌‬௜ǡ௧ା᳃௧ െ ‫݌‬௜ǡ௧
‫݌‬௜ǡ௧
ϰϱ
ϯϱ
ഓཤ 3 ᖺ࿛΅᪂Ϊঠ஦Ή ᚓ ‌῟୺ᡂศ‷‱†Ῠᰴ౯⁅⁺‽
‒⏝ῄῦḟ΅ᅇᖐᘧ‒᥎ᐃῙ​⃺
ଵହ
‫ݎ‬௜ǡ௧ ൌ ܽ௜ǡ଴ ൅ ෍ ܽ௜ǡ௞ ‫ݔ‬௞ǡ௧
௞ୀଵ
ូ ‫ݔ‬௞ǡ௧ Ῠ`᫬้‫ݐ‬ΎῊῑ​˜݇୺ᡂศ΅୺ᡂศ‷‱†ῧῂ​⃺ᅇᖐศ
ᯒ΅ၯ⃸AIC ᇶ‽[Akaike 74]ΎῊῑ​‷⁄⁁⁕•‸വᢥ‒ࢌῄ⃸
਍᫂ຊ΅పῄ਍᫂ኚᩘ`ᅇᖐᘧΎྵ ῦῄῪῄ⃺
ូ 1 ⁴᭶ᚋ΅ண ΎῊῄῦ`⃸ছɾᮇ࿛΅ฌᅇᖐศᯒ΅Ҝ⏤ᗘਚ
ᩚ῭῿ỴᐃῨእᤄᮇ࿛΅ኀؐṇµ⋡Ύṇ΅┦࿥ῌ॒ ‌῟⃺῝΅
ᵝᏊ‒ᅗ 2 Ύ♧Ῑ⃺ᅗ୰΅┤ɪ`୍ḟ΅ಶఝ᭤ɪῧῂ​⃺
ᅗ 2℆ូ ண ṇµ⋡ῨҜ⏤ᗘਚᩚ῭῿Ỵᐃಀᩘ΅┦࿥
3. እᤄண Ȇᯝ
2 ÷΅ᡭἲ‒⏝ῄῦእᤄண ΅Ɲᗘ᳨হ‒ࢌΰ῟⃺ᅇᖐᘧ΅᥎
ᐃῊ ῳእᤄண ΅ᡭᅠ`௨ୗ΅೫ ῧῂ​⃺
1998 ᖺ 1 ᭶ 1 ᪥~2000 ᖺ 12 ᭶ 31 ᪥΅ 3 ᖺ࿛(36 ⁴᭶)΅᪂
Ϊঠ஦ῧ਍᫂ኚᩘ΅ছɾ⁅⁺‽‒సᡂῙ​⃺
ኀ ؐ
ṇ
µ ⋡
1.
2.
3.
4.
ྛ᭶΅਍᫂ኚᩘῧ͓᭶ᮎ΅Ǻ್‒ࣅ਍᫂ኚᩘῨῙ​ฌᅇᖐ
ᘧ‒᥎ᐃῙ​⃺῟ῠῗ⃸ΐ΅᫬᥎ᐃῙ​΅` 2.3 ÷ῧ♧ῗ῟⁨
‽⁺ⁱ΅್ῧῂ​⃺
᥎ᐃ῕‌῟ᅇᖐᘧΎ┤ಶ΅⁄‫⁅⁆‷‫‬⁺‽(2001 ᖺ 1 ᭶ 1 ᪥
~2001 ᖺ 1 ᭶ 31 ᪥)Ή ᚓ ‌῟୺ᡂศ‷‱†‒௦ධῙ​ΐῨ
ῧ͓᭶ᮎ(2001 ᖺ 2 ᭶ᮎ)΅Ǻ್΅᥎ᐃ‒ࢌῆ⃺
ছɾ⁅⁺‽΅సᡂ࿔ጞ᪥‒ 1 ⁴᭶Ὶῤഌ Ί​ΐῨῧᅇᖐᘧ
‒ẖ᭶᭦᪂ῗῦῄ῍⃸2010 ᖺ 10 ᭶ᮎ῾ῧ΅᥎ᐃ‒ࢌῆ⃺
ϯϵ͘ϯ
ϰϬ
໬Ꮫ
๢໠៊ჰ๢
ᶵᲔ
Ҝືత
Ⴑᶵ៊Ɲᐦ
་۶៊⁖⁩‷ †
ᇇရ
ᐙᗞ⏝ရ
ၟ♫
ᑠ኎ ″⁺⁑‷
※⁓⁆․‥†
⁅‡†
೫ಙ
ᘓয
ఫᏯ៊୙ື⏘
ഐ౪
බ┈
ฐࠕ
TOPIX
᪥Ȃᖹᆒ
΅༢ਂ‒ྵ ΅`lj 10 ୓ǼᏑᅾῗ⃸࿸್௨ୖ΅ඹୣ࿥ಀ΅Ǽ
ྜΊ` 400 Ή 500 Ǽῧῂΰ῟⃺
ᜓ
᪥Ȃᖹᆒ
ᇇရ
ၟ♫
ᘓয
බ┈ ་۶࣊ࣝࢫ
723,;
ฐࠕ ႱᶵƝᐦ
ᶵᲔ ࢯࣇࢺ࢙࢘
ࢧ࣮ࣅࢫ
ఫᏯ୙ື⏘
࣓ࢹ࢕࢔
๢໠ჰ๢
ഐ౪໬Ꮫ
ᐙᗞ⏝ရ
Ҝືత
ᑠ኎ࡾ
೫ಙ
Ҝ⏤ᗘਚᩚ῭ࡳỴᐃಀᩘ
Ҝ⏤ᗘਚᩚ῭῿ỴᐃಀᩘῨ`ᅇᖐᘧ΅ᙜῦ`῾ ᗘ‒ࢢῙᣦᶆ
ῧῂ ⃸ᅇᖐᘧῌഓཤ΅ኚື‒ ῏਍᫂ῧ῍ῦῄ​ΏῩኀؐṇµ⋡
ῌ዇῏Ὺ​ഴྥῌ॒ ‌῟⃺ΐ‌`ছɾᮇ࿛ῧᙜῦ`῾ ΅ ῄᅇ
ᖐᘧ‒సᡂῧ῍ῦῄ​ΏῩእᤄண ΅Ɲᗘ ዇῾​ῨῄῆΐῨῧῂ ⃸
-2-
The 26th Annual Conference of the Japanese Society for Artificial Intelligence, 2012
ෆᤄ΅ẁၥῧእᤄண ΅Ɲᗘ‒኱῾ΉΎ᥎ῗฎ​ΐῨῌῧ῍​⃺῾
῟⃸ᅗ2Ή ᫂ ΉῪ ῆΎ⃸ᮏ◊"ῧ`ഓ๫പྜῌ⏕ῘῦῄῪῄ⃺
ഓ๫പྜῨ`⃸ছɾᮇ࿛ῧᙜῦ`῾ ΅዇ῄ⁅⁺‽‒సᡂῙ​ῂ῾
Ύỗ⏝ᛶΎḞῑ⃸ᮍ᮶΅⁅⁺‽΅ண Ύ`പ῕ῪῄᙧῨῪΰῦῗ
῾ῆΐῨῧῂ​⃺᪂ΪῨῄῆȂ῭ືྥΎ࿥Ῑ​ɉ̦ⓗῪ᝟ሗ‒ᢅῆ⁄
‫⁅⁆‷‫‬⁺‽Ή ⃸୺ᡂศศᯒΎ ΰῦྜᡂኚᩘ‒సᡂῗ⃸῕ Ύᅇ
ᖐศᯒ΅ၯΎ‷⁄⁁⁕•‸വᢥ‒⏝ῄ​ΐῨῧഓ๫പྜῌᅇഺ῕
‌῟ῨͳῈ ‌​⃺
ូ ୍᪉ῧ⃸೫ಙ බ┈ ᘓয⃸་⒪⁹⁖⁩‷ †Ύ࿥ῗῦ`Ҝ⏤ᗘ
ਚᩚ῭῿Ỵᐃಀᩘ⃸እᤄண ΅ኀؐṇµ⋡Ῠ ΎపῄỈ‽ῧῂΰ
῟⃺ΐ‌ ΅⏘ᴗῌෆႲ⏘ᴗῨ࿧῰‌⃸᪂Ϊঠ஦Ή ΐ‌ ΅⏘ᴗ
Ύ࿥Ῑ​≉ᚩฎῌῆ῾῏ᢳฟῧ῍ῪΉΰ῟῟ ῧῂ​ῨͳῈ ‌​⃺
ូ ᐇၯΎᢳฟ῕‌῟ඹୣ༢ਂ΅⁘†‒॒ῦ῿​Ῠ⃸ᾲ࿇ᮇฐ฼᫢ୖ
᪼ᾳ ᾲ⁅⁓᫢ඞ᭹ᾳῪῩῨῄΰ῟Ȃ῭඲యΎస⏝Ῑ​ῨͳῈ ‌​
ඹୣ⃸ᾲůᅜ᫢⁆⁦‽Ҝືతᾳ ᾲఫᏯ᫢ࠕଘᾳ΅ ῆΎಶู΅ᴗ✀Ύᙳ
ᅚ‒ཬῼῙῨͳῈ ‌​ඹୣῌ॒ ‌῟⃺ΐ‌ ΅ฟ⌧⁏‽⁺ⁱ‒
⏝ῄῦᅇᖐᘧ‒᥎ᐃῗῦῄ​῟ ⃸≉ᚩฎῌῆ῾῏ᢳฟῧ῍​ᴗ✀
Ῠ῝ῆῧῪῄᴗ✀ῌᏑᅾῙ​⃺῝‌ῌᅗ 2 ΅Ҝ⏤ᗘਚᩚ῭῿Ỵᐃ
ಀᩘΎࢢ‌ῦῄ​⃺ῗΉῗ⃸⁄‫⁆‷‫‬Ῠῄῆே࿛ῌॾจ‒ࢌῆΐῨ΅ῧ
῍​᝟ሗ‒⏝ῄῦᰴ౯ண ‒ࢌΰῦῄ​῟ ⃸ྛᾫ΅᥎ᐃΎᑐῗῦ
Ῡ΅ ῆῪ᝟ሗῌᐇၯΎຠᯝ‒ᣢΰῦῄ῟΅Ή‒ख़ड़ⓗΎ☜Ή ​ΐῨῌῧ῍​⃺ΐ‌`ഓཤ΅ኚືΎᑐῗῦศᯒῌྍЏῧῂ​஦⃸
῝ῗῦᮍ᮶΅ᰴ౯΅᥎ᐃ΅ၯΎ⏝ῄ​ΐῨ΅ῧ῍​ྍЏᛶ‒♧ῗῦ
ῄ​⃺
3.2 ண ᮇ࿛ู⁹዇ኚືᮇ΅ண ṇµ⋡
3.1 ÷ῧ` 1 ⁴᭶ᚋ΅እᤄண Ύῤῄῦ೉Ό῟ῌ⃸ᮏᡭἲΎ ΰ
ῦῩ΅⛬ᗘඛ΅ᮍ᮶῾ῧ᥎ᐃῧ῍​΅Ή‒᳨হῗ῟⃺ᮇ࿛ู΅እ
ᤄண ΅ኀؐṇµ⋡‒ᅗ 3 ᕥ΅ᩳɪᲬ‫⁓⁧‮‬ῧ♧Ῑ⃺ᮇ࿛ูῨ`
ண ᮇ࿛΅࿇῕ῧῂ ⃸2 ÷ῧ೉Ό῟ ῆΎ 1 ⁴᭶ᚋ⃸2 ⁴᭶ᚋ⃸3 ⁴
᭶ᚋ΅ 3 ῤ΅ᮇ࿛Ύῤῄῦண Ɲᗘ‒᳨হῗ῟⃺
ូ ண ᮇ࿛ῌ࿇῏Ὺ​Ύῤ‌⃸῝΅࿛Ύୣΐ​஦૏΅ᩘῌቑῈ​῟
⁄‫⁆‷‫‬ῨȂ῭ኚື΅┦࿥`ᙅ῾​⃺ᅇᖐᘧΎ⏝ῄ ‌῟୺ᡂศ
΅॒ῦ῿​Ῠ 2 ⁴᭶ᚋ⁹3 ⁴᭶ᚋ΅ண ῧ`⃸1 ⁴᭶ᚋ΅ண ΅ 1.5
ಸΉ 2.0ಸ΅ᩘ΅୺ᡂศῌ⏝ῄ ‌ῦῄ῟⃺ΐ‌`਍᫂ຊ΅ᙅῄ
਍᫂ኚᩘ‒ከ῏⏝ῄ​ΐῨῧȂ῭ኚືΎྜҤῙ​ᅇᖐᘧ‒సᡂῗῦ
ῄ῟ΐῨΎῪ ⃸ഓཤ΅ኚື΅਍᫂`ῧ῍ῦ ᮍ᮶΅ኚືΎᑐῙ​
ண ຊῌؐῡ​⃺ᐇၯΎ⃸ᮇ࿛ῌ࿇῏Ὺ​Ύῤ‌ῦண Ɲᗘῌᝏ῏
Ὺ​ഴྥῌ॒ ‌῟⃺೛Ύ⃸1 ⁴᭶ᚋ΅ண Ύ࿥ῗῦ 2 ⁴᭶ᚋ⃸3 ⁴
᭶ᚋ΅ண ᑡῪῄ୺ᡂศΎ ​᥎ᐃῧዲᡂʬ‒཰ ῟Ῠῄῆ
ΐῨ`⃸῝‌ῠῑȂ῭΅ኚືΎ┤Ȇῗ῟୺ᡂศ‒సᡂῧ῍ῦῄ῟Ῠ
ῄῆΐῨῧῂ​⃺
ϲϬ
ኀ
ؐ
ṇ
µ
⋡
⃲
ϱϴ
ϱϲ
ண ᮇ࿛ῌ࿇῏Ὺ​Ύῤ‌ῦኀؐṇµ⋡ῌୗῌ​୍᪉ῧ⃸ኚື
΅኱῍Ὺ᫬ᮇΎῊῑ​ண ṇµ⋡ῧ` 1 ⁴᭶ᚋ⁹2 ⁴᭶ᚋ΅ண ῧ
2 ⁛•ⁱ⁆௨ୖ΅ᨵၿῌ॒ ‌῟(ᅗ3 ྑ΅ᓞᲬ‫⃺)⁓⁧‮‬ΐ΅ΐῨΉ ᮏᡭἲ` 2 ⁴᭶ᚋ΅ண ῾ῧ᭷ຠῧῂ​ῨῄῆΐῨῌῧ῍​⃺ῪῊ⃸
዇ኚືᮇ΅ண Ῠ`ୗᘧΎ ΰῦᐃ῾​⃸᫬้ T Ή ጞ῾​ছɾᮇ
࿛ 36 ⁴᭶ΎῊῑ​⁨‽⁺ⁱ΅ᶆ‽೫ᕪ ߪ ΅ 0.1 ಸ ȟᑐ್ῌ኱
῍Ὺኚື‒᥎ᐃῗ῟ሙྜ΅῿ண ‒ࢌΰ῟ ΅ῧῂ​⃺
்ାଷହ
ଶ
ߪ ଶ ൌ ෍ ሺ‫ݎ‬௜ǡ௧ െ ‫ݎ‬തതതതሻ
పǡ௧
௧ୀ்
ΐ΅࿸್Ύ ΰῦᐃ ῟዇ኚືᮇ`඲ண ᮇ࿛΅Ὴ ῝ 3 ศ΅ 1
ῧῂΰ῟⃺ண ῗ῟ᅇᩘ΅ῆῡ⃸ኀؐῌ୍Ҥῗ῟๭ྜ‒ᅗ 3 ΅ᓞῄ
Წ‫⁓⁧‮‬ῧ♧ῗῦῄ​⃺
4. ᥦ᱌ᡭἲ΅‵‷⁄⁞໬
ᐇၯ΅ฐࠕ࿥ಀ͵΅౑⏝‒┠ⓗῨῗῦ⃸ᮏ◊"ῧᥦ᱌ῗ῟ᡭἲ
΅‵‷⁄⁞໬‒ࢌΰ῟⃺3 ÷῾ῧ΅ᐇቲῧᢞଘᐙΉ ोᮃ΅዇ῄ࿇
ᮇண ΅᭷ຠᛶ‒☜Ή ῟⃺῝΅ୖῧᢞଘᐙῌ෇⁥Ύᢞଘ‒ࢌῆ
῟ ΅ॾจ‒ࢌῄ⃸ഓཤ΅ኚືΎᑐῗῦ ศᯒῧ῍​‵‷⁄⁞‒ᥦ
᱌Ῑ​⃺ഓཤ΅ᰴ౯ኚື ⁄‫΅⁆‷‫‬᝟ሗ‒ᥦ♧Ῑ​΅῿ῧῪ῏⃸
ศᯒΎᇶῥῄ῟᝟ሗ‒ᥦ౪Ῑ​⃺
4.1 ᥦ᱌‵‷⁄⁞ᴫो
ᮏ‵‷⁄⁞` forecast ‽⁔⃸tag cloud ‽⁔⃸ඹୣ࿥ಀฟ⌧⁏‽⁺
ⁱ୍।ῧᵓᡂ῕‌ῦῄ​⃺ᅗ 4 ῌ forecast ‽⁔⃸ᅗ 5 ῌ tag cloud ‽
⁔⃸ᅗ 6 ῌඹୣ࿥ಀฟ⌧⁏‽⁺ⁱ୍।ῧῂ​⃺ᮏ÷ῧ`῝‌῞‌
΅ᅗΎᑐῗῦᶵЏ‒਍᫂Ῑ​⃺
ṯforecast ‽⁔
ᣦᶆ῔Ῠ΅ண ΅ኀؐ‒ᰴ౯ኳẼணሗ΅ᙧῧᅗ 4 ᕥ΅ࢢΎ῾Ῠ ῦῄ​⃺ᅠΎ 1 ⁴᭶ᚋ⃸2 ⁴᭶ᚋ⃸3 ⁴᭶ᚋ΅ண ‒ࢢ♧ῗ῟ ΅ῧ
ῂ ⃸୍┠ῧȂ῭ືྥῌᢕᥱῧ῍​ ῆΎῪΰῦῄ​⃺ᅗ 4 ྑ΅‫⁧‮‬
⁓`ᣦᶆ῔ῨΎ‽⁔ῧษ ᭰Ὲ ‌​ ῆΎῪΰῦῊ ⃸ᐇၯ΅⁨‽
⁺ⁱῨᅇᖐῧ᥎ᐃ῕‌῟⁨‽⁺ⁱ‒ছɾᮇ࿛΅ 36 ⁴᭶+ண ΅ 1 ⁴
᭶ศࢢ♧ῗῦῄ​⃺⁤⁺‴`‫ୖ⁓⁧‮‬ൺΎῂ​ a Ή u ΅‽⁔‒ษ ᭰Ὲ​ΐῨῧಶู΅ᣦᶆ‒ศᯒῙ​ΐῨ ྍЏῧῂ​⃺῾῟⃸‫⁓⁧‮‬
ୖ΅ྛⅬ‒‭⁨⁁‭Ῑ​Ῠ⁛⁁⁕†⁁⁕ῌࢢ♧῕‌⃸ྛ᫬ⅬΎῊῄῦ
ኚືΎ᭱ ᐤ୚ῗ῟୺ᡂศῌࢢ♧῕‌​⃺ኚືΎ᭱ ᐤ୚ῗ῟୺ᡂ
ศῨ`⃸῝΅᫬ⅬΎῊῄῦᅇᖐಀᩘῨ୺ᡂศ‷‱†΅
΅ȟᑐ್
ῌ᭱ ኱῍ῄ୺ᡂศ΅ΐῨῧῂ​⃺pre/next ΅⁨ⁱ‭‒‭⁨⁁‭Ῑ​ΐῨ
ῧ๓᭶/͓᭶΅ศᯒȆᯝΎ⛣ືῙ​⃺
ϱϴ͘ϰ
඲ᮇ࿛
ϱϲ͘ϭ
ϱϯ͘ϱ
ϱϰ
ϱϭ͘Ϯ
ϱϮ
ϱϬ
዇ኚືᮇ
ϱϬ͘ϱ
ϰϴ͘ϴ
ϰϴ
ϰϲ
ϰϰ
ϭ⁴᭶ᚋ
Ϯ⁴᭶ᚋ
ϯ⁴᭶ᚋ
ண ᮇ࿛
ᅗ 3℆ូ ண ᮇ࿛ูண ṇµ⋡⁹዇ኚືᮇ΅ண ṇµ⋡
ᅗ℀℆ូ forecast ‽⁔
-3-
The 26th Annual Conference of the Japanese Society for Artificial Intelligence, 2012
ṯtag cloud ‽⁔
ΐ΅᫬ᮇ΅ᅇᖐᘧ΅਍᫂ኚᩘῨῪΰ῟ྛ୺ᡂศΎᑐῗῦ⃸ᅉᏊ૷
֤ฎ΅኱῍Ὺ༢ਂ‒ྵ ඹୣ༢ਂ⁘†΅୍౛‒୺ᡂศ῔ῨΎ♧ῗ
ῦῄ​⃺ᅗ 4 ΅‫⁕⁁†⁕⁁⁛΅⁓⁧‮‬ῧࢢ♧῕‌​ᙳᅚຊ΅኱῍Ὺ
୺ᡂศῨ⃸ΐ΅ࢢ‒॒ẚΌ​ΐῨῧ⃸Ῡ΅ ῆῪ༢ਂῌῩ΅ᣦᶆ΅Ῡ
΅ኚືΎຠῄῦῄ῟΅ΉਚΌ​ΐῨῌῧ῍​⃺῾῟⃸ྛ୺ᡂศ΅୺
ोῪඹୣ༢ਂ‒୍।ῧ῾Ῠ ῦῄ​΅`ഓཤ΅ศᯒΎ ᑐᛂῙ​
῟ ῧῂ ⃸ΐΐ‒୍॒Ῑ‌῰Ῡ΅ ῆῪȂ῭࿥ೳ΅ঠ஦ῌῂΰ῟
΅Ή‒኱῾ΉΎᢕᥱῙ​ΐῨῌῧ῍​⃺῾῟⃸ྛ୺ᡂศྡ⃴PC1⃸
PC2⃸…⃸PC15⃵`⁨ⁱ‭ᵓ೰‒ᣢΰῦῊ ⃸‭⁨⁁‭Ῑ​ΐῨῧᅗ 6 ΅
ඹୣ࿥ಀฟ⌧⁏‽⁺ⁱ‒॒​ΐῨῌῧ῍​⃺
ᑠ῕ῄ⃺῝΅୍᪉ῧᑡῪῄ୺ᡂศ΅ᩘῧኚື‒਍᫂ῧ῍ῦῄ​῟
⃸ᅗ 5 ᅗ 6 ΅୺ᡂศ୰΅ᅉᏊ૷֤ฎῌ኱῍ῄ༢ਂ‒ྵ ඹୣ
⁘†ῌῩ΅ ῆΎᰴ౯΅ኀؐΎᙳᅚῗῦῄ​΅Ή‒ᢕᥱῧ῍​⃺
ᮏ‵‷⁄⁞ῧ`ഓཤ΅ᰴ౯⁅⁺‽Ῠ⁄‫⁅⁆‷‫‬⁺‽΅┦࿥ᛶ‒ྍ
ख़໬Ῑ​ΐῨῧ⃸ॾจ‒ྍЏΎῗῦῄ​⃺ഓཤ΅┦࿥Ή ᮍ᮶΅ண
΅ॾจΎᙺWῦ​ΐῨῌᮏ‵‷⁄⁞΅┠ⓗῧῂ​⃺
5. ῾Ῠ ᮏ◊"ῧ` CPR ἲ‒ᛂ⏝ῗῦlj 10 ᖺ࿛Ῠῄῆ࿇ῄ‷⁏ⁱῧ΅እ
ᤄண ‒ࢌΰ῟Ȇᯝ⃸1 ⁴᭶ᚋ΅ண ΎῊῄῦ TOPIX ᪥Ȃᖹᆒ
Ῠῄΰ῟ᕷሙᖹᆒᰴ౯ῧ` 60%௨ୖ΅ኀؐṇµ⋡‒཰ ῟⃺῾῟⃸
ᴗ✀ูᖹᆒᰴ౯‒ྵ ῟ኀؐṇµ⋡΅ᖹᆒῧ 55%௨ୖῨῄῆ዇
ῄƝᗘ‒♧ῗ῟⃺ΐ‌`ᐇົῧồ ‌​Ỉ‽΅ṇµ⋡ῧῂ​[7]⃺
῕ Ύ⃸ᮏ◊"ῧ⏝ῄ῟ᡭἲ‒ᇶΎ⃸ᢞଘᐙῌॾจ‒ࢌῆΐῨ‒๓
ᥦῨῗ῟Ȃ῭ືྥศᯒ‵‷⁄⁞ ᵓÿῗ῟⃺ΐ΅‵‷⁄⁞Ύ ⃸௒
῾ῧ୙᫂░ῧῂΰ῟Ȃ῭ኚືῨ⁄‫⁆‷‫‬᝟ሗῨ΅࿥ಀᛶῌ᫂ ΉῨ
Ὺΰ῟⃺ᢞଘᐙ`ᮏ‵‷⁄⁞‒⏝ῄῦॾจ‒ࢌῄ⃸ᢞଘάືΎᙺ
Wῦ​ΐῨῌྍЏῧῂ​⃺
ូ ᮏ◊"ῧ⏝ῄ῟ᡭἲΎ`῾ῠᨵၿ΅వᆅῌῂ​⃺౛Ὲ῰ᙧែǢ
ॾᯒ΅ẁၥῧ ೳਂῌศॾ῕‌ῦῗ῾ΰῦῄ​Ⅼ ⃸኱ᐜฎ΅⁅
⁺‽Ῠῗῦ᪂Ϊঠ஦΅⁅⁺‽ῠῑΎးᐃῗῦῗ῾ΰῦῄ​Ⅼῧῂ​⃺
ᐇၯ΅ᢞଘΎ`ᵝᾫῪ᝟ሗ‒⏝ῄῦॾจ‒ࢌῆ῟ ⃸༢୍΅᝟ሗ
ῧ`Ὺ῏ࣽᩘᣦᶆΎᑐᛂῗ῟‵‷⁄⁞‒ᥦ᱌ῗῦῄ῍῟ῄ⃺
ཧͳᩥ⊩
ᅗ 5℆ូ tag cloud ‽⁔
ṯඹୣ࿥ಀฟ⌧⁏‽⁺ⁱ୍।
ছɾᮇ࿛⁹እᤄᮇ࿛ΎῊῑ​୺ो༢ਂ⁘†΅᫬DŽิⓗῪฟ⌧⁏‽
⁺ⁱ᝟ሗ‒♧ῗῦῄ​⃺୺ो༢ਂῨ`ྛ୺ᡂศΎῊῄῦᅉᏊ૷֤
ฎῌୖ఩ 15 ఩῾ῧ΅ ΅Ῠῗ῟⃺῝΅୺ो༢ਂ‒ྵ ඹୣ༢ਂ⁘
†඲ῦ΅ฟ⌧⁏‽⁺ⁱ‒ᮏ‵‷⁄⁞ῧ`ࢢ♧ῗ⃸ୖΎῂ​ΏῩᅉᏊ
૷֤ฎῌ኱῍ῄ༢ਂ‒ྵ ⁘†ῨῪΰῦῄ​⃺ΐ΅ฟ⌧΅᭷↓Ῠᰴ
౯΅ኀؐ‒ྜ‫‏‬ΊῦศᯒῙ​ΐῨῧ⃸Ῡ΅ ῆῪ⁘†΅ฟ⌧΅᭷↓
ῧኀؐῌỴ῾ΰῦῄ​΅Ή‒ᢕᥱῙ​ΐῨῌῧ῍​⃺
[AERA] AERA’2012.2.13⃸pp.62⃸ᮅ᪥᪂Ϊฟ∧.
[Akaike 74] Akaike, H.: A new lool at the statistical model
identification, IEEE Transactions on Automatic Control, Vol.19,
pp.716-723 (1974).
[ChaS] ChaSen ⁙⁺⁞⁘⁺‶ : http://chasen.naist.jp/hiki/chasen .
[዇ 2002] ዇ὒ⃸֘஭ၟࢌ⃸኱yᩒ⃸⏣୰࢘ ℆ ⁉⁣⁺⁧⁩⁋
⁁⁆⁺‭Ύ ​᫬ᮇ΅ᰴ౯ண ⃹ᰴ౯ண ΎῊῑ​⁓‡⁩‽⁨ⁱ‫‮‬
Ύ ​≉ᚩฎᢳฟ⃹⃸ႱᏊ᝟ሗ೫ಙᏛ఍ᢏ࢑◊"ሗ(NPL)⃸
Vol.102⃸No.432⃸ pp.13-16 (2002) .
[ఀᗞ 2006] ఀᗞᩧᚿ ℆ ೽໬ਭⓗᡭἲ‒⏝ῄ῟ฐࠕ⁅⁺‽΅ண
⃸᪥ᮏಙᆇᛶᏛ఍৵⃸Vol.28⃸No.7⃸pp.471-480 (2006) .
[࿴Ἠ 2011] ࿴Ἠ₩⃸ᚋ‫ܘ‬༟⃸ᯇ஭‫ܘ‬஬൫℆ Ȃ῭⁄‫⁆‷‫‬᝟ሗ‒⏝
ῄ῟࿇ᮇⓗῪᕷሙືྥ᥎ᐃ⃸᝟ሗฎ⌮Ꮫ఍৵⃸Vol.52⃸No.12⃸
pp.3309-3315 (2011) .
[኱⃝ 2006] ኱⃝ᖾ⏕ : ‿⁡ⁱ‷Ⓨ॒΅⁅⁺‽ศᯒ᫢⁠⁅⁩໬+
ྍख़໬+‱⁝⁣⁉ ⁺‵⁥ⁱ᮫‵⁈⁨
๰Ⓨ⃸ᮾிႱᶵ኱Ꮫฟ∧ᒁ
(2006).
ᅗ 6℆ូ ඹୣ࿥ಀฟ⌧⁏‽⁺ⁱ୍।
4.2 ᮏ‵‷⁄⁞‒⏝ῄ῟Ȃ῭ືྥศᯒ΅฼⏝౛
᭱ึΎ forecast‽⁔΅ᬕ/Ⴅ΅ࢢ୍।ῧ௒ᚋ΅Ȃ῭⁆ⁱ⁇‒୍┠
ῧᢕᥱῧ῍​ ῆΎῪΰῦῄ​⃺‫⁓⁧‮‬Ύῤῄῦ`ഓཤ΅ኚືΎᑐ
ῗῦෆᤄ࿥ᩘ‒సΰῦῄ​ῌ⃸Ҝ⏤ᗘਚᩚ῭῿Ỵᐃಀᩘ`ẚుⓗ
[ ᪥Ȃ] ᪥Ȃ‵ ※ ⁺ ⁧ ‷ ⃸ ᪥ᮏȂ῭᪂Ϊ⁅ ‶ ‽ ⁩ ⁅ ‡ †
http://vip-test2.nikkei.co.jp/help/contract/price/02/help_KIJI_thes.html
[ญᮧ] ญᮧ੷ๆฐࠕᕤᏛ◊"‹ⁱ‽⁺⁙⁺⁞⁘⁺‶ : http://
qr.nomura.co.jp/jp/n40/index.htmlᬌᬌ
ᬌᬌ῟ῠῗ⃸12085$ ΅ᥦ౪` ᖺ ᭶ᮎ᪥‒ ΰῦ
Ǻ஢ῗῦῄ῾Ῑ⃺ូ
-4-