Title
進化計算を用いた複数ユーザに好まれる香りの探索－LANを介し
たシステムの構築－
Author(s)
福本 誠, 原 大海
Citation
福岡工業大学総合研究機構研究所所報 第2巻 P73-P77
Issue Date
2020-2
URI
http://hdl.handle.net/11478/1485
Right
Type
Departmental Bulletin Paper
Textversion
Publisher
福岡工業大学 機関リポジトリ
FITREPO
⚟ᒸᕤᴗᏛ◊✲ᡤᡤሗ㸦2019㸧
㐍ィ⟬ࢆ⏝࠸ࡓ」ᩘ࣮ࣘࢨዲࡲࢀࡿ㤶ࡾࡢ᥈⣴
㸫/$1 ࢆࡋࡓࢩࢫࢸ࣒ࡢᵓ⠏㸫
⚟ᮏ ㄔ㸦ሗᕤᏛ㒊ሗᕤᏛ⛉㸧
ཎ ᾏ㸦Ꮫ㝔ᕤᏛ◊✲⛉ሗᕤᏛᑓᨷ㸧
The Search of the Fragrance Preferred by Multiple Users based on Evolutionary Computation
㸫
Construction of the Concrete System via LAN㸫
Makoto FUKUMOTO㸦Department of Computer Science and Engineering, Faculty of Information Engineering㸧 Hiromi HARA㸦Computer Science and Engineering*UDGXDWH6FKRRORI(QJLQHHULQJ㸧
Abstract
Interactive Evolutionary Computation (IEC) is well known approach searching good solutions for each user in terms of finding good graphics, sounds, and music pieces. This study aims to propose IEC with multiple users for creating fragrance. The fragrances are composed of several aroma sources; therefore, target of the proposed method is to optimize intensity of each source aroma. While the general IEC searches good solutions for each of the users, the proposed method can search good or optimal solution between the users by changing good solutions between the users during the search. In this study, a concrete system was constructed to show a fundamental efficiency of the proposed method.
Keywords㸸Interactive Evolutionary Computation, Multiple Users, Fragrance
1. ࡣࡌࡵ
㏆ᖺ㸪㤶ࡾࡣ㤶Ỉࡸ࣐ࣟࢸࣛࣆ࣮ࢆጞࡵ㸪᪥ᖖ࡛⏝ ࡉࢀࡿ〇ရࡢ㤶ࡾ࡙ࡅ࡞㸪ᵝࠎ࡞⏝㏵࡛⏝ࡉࢀ࡚࠸ ࡿ㸬ࡑࡢࡓࡵ㸪௨๓ࡼࡾࡶ㌟㏆࡞ࡶࡢ࡞ࡗ࡚࠸ࡿࡢ࡛ࡣ ࡞ࢁ࠺㸬ᗄࡘࡢࢹࣂࢫࡣ㸪ࢥࣥࣆ࣮ࣗࢱ᥋⥆ࡍ ࡿࡇ࡛ᡭ㍍㤶ࡾࢆ࣮ࣘࢨᥦ౪ࡍࡿࡇࢆྍ⬟ࡋ㸪 ࡉࡽࡣ」ᩘࡢ㤶ᩱࢆΰྜࡋ࡚ᥦ♧ࡍࡿࡇࡶ࡛ࡁࡿ㸬ࡇ ࠺࠸ࡗࡓࢹࣂࢫࡣ㸪ᚑ᮶࡞ࡗࡓ㤶ࡾࡢ⏝ࡸᴦࡋࡳ ᪉ࡘ࡞ࡀࡿ⪃࠼ࡽࢀࡿ㸬 ࡋࡋ࡞ࡀࡽ㸪ึᚰ⪅ࡢ࣮ࣘࢨࡀᡭసᴗ࡛㤶ᩱࢆΰࡐ㸪 ዲࡳྜ࠺㤶ࡾࢆ⏕ᡂࡍࡿࡇࡣᅔ㞴࡛࠶ࡿࡓࡵ㸪ᴗ࡞ ࡢ〇㐀⪅ഃࡽᥦ౪ࡉࢀࡿࡶࡢࢆ⏝ࡍࡿࡢࡀ୍⯡ⓗ࡛ ࠶ࡿ㸬ࡲࡓ㸪࣐ࣟࣞࢩࣆ1)ࡢࡼ࠺࡞ᙧ࡛㤶ࡾࡢㄪྜẚࡢ ࡀᥦ౪ࡉࢀ࡚࠸ࡿࡶࡢࡢ㸪ಶேᕪࡀ࠶ࡾከᵝ࡛࠶ࢁ࠺ಶே ࡢឤᛶࡘ࠸࡚⪃࠼ࡿ㸪ᥦ౪ࡉࢀࡿࡶࡢࡀᮏᙜࡢព࡛ ಶࠎࡢ࣮ࣘࢨࡢዲࡳྜ࠺ࡶࡢ࡛࠶ࡿࡣゝ࠸㞴࠸㸬ࡢ ࣓ࢹࡘ࠸࡚ࡶゝ࠼ࡿࡇ࡛࠶ࡿࡀ㸪ࡇ࠺࠸ࡗࡓၥ㢟 ࡣ㸪࣮ࣘࢨࡢឤᛶࡀࣈࣛࢵࢡ࣎ࢵࢡࢫⓗ࡞≉ᛶࢆᣢࡘࡇ 㸪࣓ࢹࡢㄪᩚࡀᅔ㞴࡛࠶ࡿࡇࡢ 2 ࡘࡢせᅉࡼࡿ ࡶࡢゝ࠼ࡿ㸬 ಶࠎࡢ࣮ࣘࢨྜ࠺࣓ࢹࢥࣥࢸࣥࢶࢆ᥈ࡋฟࡍᡭἲ ࡢ୍ࡘࡋ࡚㸪ᑐヰᆺ㐍ィ⟬㸦Interactive Evolutionary Computation㸧2,3)ࡀ▱ࡽࢀ࡚࠸ࡿ㸬ࡇࢀࡣ㸪㑇ఏⓗࣝࢦ ࣜࢬ࣒ࢆࡣࡌࡵࡍࡿ㐍ィ⟬㸪ホ౯㛵ᩘࡋ࡚࣮ࣘࢨ ࡢឤᛶࢆᑟධࡋࡓᡭἲ࡛࠶ࡿ㸬ࡍ࡞ࢃࡕ㸪᭱▷⤒㊰᥈⣴ࡢ ၥ㢟࠾ࡅࡿ㊥㞳ࢆィ⟬ࡍࡿホ౯㛵ᩘࡢᙺࢆ㸪࣮ࣘࢨ⮬ ㌟ࡀᢸ࠺࠸࠺ࡶࡢ࡛࠶ࡿ㸬↓ㄽ㸪ࡇࡢሙྜࡣ㸪࣮ࣘࢨ ࠼ࡽࢀࡿࡢࡣ⤒㊰࡛ࡣ↓ࡃ㸪࣓ࢹࢥࣥࢸࣥࢶࡢࡼ࠺ ࡞่⃭࡛࠶ࡾ㸪ᚑ᮶ࡣどぬࡸ⫈ぬ㛵ࡍࡿࡶࡢࡀࢇ ࡛࠶ࡗࡓ3)㸬 ࡇࡇ࡛ࡣ㸪㤶ࡾࡢΰྜࢆᑐ㇟ࡍࡿᑐヰᆺ㐍ィ⟬ࡘ ࠸࡚㸪」ᩘ࣮ࣘࢨࡀཧຍࡍࡿᡭἲࢆᥦࡍࡿ㸬ࡘࡲࡾ㸪ಶࠎ ࡢ࣮ࣘࢨྜ࠺㤶ࡾࡢ᥈⣴ࡽ㸪」ᩘࡢ࣮ࣘࢨࡀዲࡴ㤶ࡾ ࡢ᥈⣴ᒎ㛤ࡍࡿ㸬ᑐヰᆺ㐍ィ⟬ࡼࡿ㤶ࡾࡢ᥈⣴ᡭ ἲࡣ㸪ⴭ⪅ࡽࡼࡗ࡚ᥦࡉࢀࡓ㸬᭱ึࡢᡭἲ4)ࡣ㑇ఏⓗ ࣝࢦࣜࢬ࣒ࢆ⏝࠸ࡓࡶࡢ࡛࠶ࡗࡓࡀ㸪ࡼࡾ᥈⣴ᛶ⬟ࡀ㧗࠸ ࡉࢀࡿᕪศ㐍࡞ࡢࡢࣝࢦࣜࢬ࣒ࡢᑟධࡀᥦࡉ ࢀ࡚ࡁࡓ5,6)㸬ࡇࢀࡽࡣ㸪Takagi ࡽࡼࡗ࡚ᥦࡉࢀࡓᑐẚ ㍑ࡢホ౯᪉ἲ7)ࢆ᥇⏝ࡋࡓࡶࡢ࡛࠶ࡿ㸬 ࡉࡽ㸪ẚ㍑せࡋࡓ㛫ࡢሗࢆ᥈⣴ᛶ⬟ࡢྥୖ⏝ ࠸ࡿᡭἲࡶᥦࡉࢀ࡚࠸ࡿ8)㸬ࡲࡓ㸪」ᩘ࣮ࣘࢨࡀཧຍࡍࡿ ᑐヰᆺ㐍ィ⟬ࡋ࡚㸪࠸ࡃࡘࡢᡭἲࡀᥦࡉࢀ࡚࠸ࡿ 9-12)㸬ከࡃࡢ࣮ࣘࢨࡢឤᛶྜ࠺ゎࡣ㸪ಶࠎࡢ࣮ࣘࢨྜ࠺ ゎࡼࡾࡶ‶㊊ᗘࡣపࡃ࡞ࡿྍ⬟ᛶࡀ࠶ࡿࡀ㸪ඹྠ࡛⏝࠸ࡿ ࣓ࢹࡋ࡚⪃࠼ࡿ㸪ࡼࡾᗈ࠸⏝㏵ࡀᮇᚅ࡛ࡁࡿ㸬㤶⚟ᮏㄔ㸪ཎᾏ ࡾ᥈⣴ࡢศ㔝࡛ࡣ㸪㤶ࡾ㛵ࡍࡿ〇ရࡢᛂ⏝ࡀ⪃࠼ࡽࢀ ࡿ㸬 ᮏ◊✲ࡢ┠ⓗࡣ㸪ᑐヰᆺ㐍ィ⟬ࡼࡿ࣓ࢹࢥࣥࢸ ࣥࢶࡢ᭱㐺ゎ᥈⣴ࡢᢏ⾡ࢆࡶ㸪ࡇࢀࡲ࡛ᥦࡉࢀ࡚࠸ ࡞࠸」ᩘࡢ࣮ࣘࢨࡀཧຍࡋ࡚ඹ㏻ࡋ࡚ዲࡴ㤶ࡾࡢ᥈⣴ࢆ⾜ ࠺ᡭἲࢆᥦࡍࡿࡇ࠶ࡿ㸬ᮏㄽ࡛ࡣ㸪ࢩࢫࢸ࣒ᵓ ⠏║ࢆ⨨ࡃࡀ㸪ࢩࢫࢸ࣒ࡢືసࢳ࢙ࢵࢡࢆ┠ⓗ⾜ࡗ ࡓᇶ♏ⓗ࡞ᐇ㦂ࡢ⤖ᯝࡶ♧ࡍ㸬
2. ᥦᡭἲ㸸」ᩘ࣮ࣘࢨࡀཧຍࡋ࡚㤶ࡾ࣓ࢹ
ࢆ᥈⣴ࡍࡿᑐヰᆺ㐍ィ⟬
ᮏ❶࡛ࡣ㸪ᥦᡭἲࡘ࠸࡚ㄝ᫂ࡍࡿࡶ㸪㑇ఏⓗ ࣝࢦࣜࢬ࣒ࡸ㤶ࡾࡢ᭱㐺ᡭἲࡘ࠸࡚ㄝ᫂ࡍࡿ㸬 ࠑ2㺃1ࠒ ᑐヰᆺ㐍ィ⟬㑇ఏⓗࣝࢦࣜࢬ࣒ ᑐヰ ᆺ㐍ィ⟬ࡣ㸪᭱㐺ゎ᥈⣴ᡭἲ࡛࠶ࡿ㐍ィ⟬ࡢホ౯㛵ᩘ ࢆ㸪࣮ࣘࢨ⨨ࡁ࠼ࡓᡭἲ࡛࠶ࡿ㸬࣮ࣘࢨࡢឤᛶࡣ㸪ࣘ ࣮ࢨ⮬㌟࡛ࡶᢕᥱ࡛ࡁ࡞࠸ࡓࡵ㸪ᵝࠎ࡞࣓ࢹࢥࣥࢸࣥ ࢶࢆ่⃭ࡋ࡚ཷࡅྲྀࡾᚓⅬࡅ࣭㑅ᢥ࡞ࡢ᪉ἲ࡛ホ౯ ࡋ㸪ࡑࡢホ౯ࢆࡶ᥈⣴ࢆ㐍ࡵࡿⅬᑐヰᆺ㐍ィ⟬ࡢ ≉ᚩࡀ࠶ࡿ㸬 ᑐヰᆺ㐍ィ⟬࡛㢖⦾⏝࠸ࡽࢀࡿ㐍ィ⟬ࡢࣝࢦࣜ ࢬ࣒ࡣ㸪㑇ఏⓗࣝࢦࣜࢬ࣒࡛࠶ࡿ㸬ࡑࡢฎ⌮ࡢὶࢀࢆᅗ1 ♧ࡍ㸬୍⯡ⓗࡣ㸪ಶయ㞟ᅋࡢ⏕ᡂ㸪ಶయࡢホ౯㸪㑅ᢥ㸪 ཫ✺↛ኚ␗㸪࠸࠺ฎ⌮ࡽ࡞ࡿ㸬ࡇࡇ࡛㸪ಶయ࠸ ࠺ࡢࡣゎೃ⿵ࢆᣦࡋ㸪㏻ᖖࡣDḟඖࡢኚᩘࡽ࡞ࡿ㸬ࡲࡓ㸪 ホ౯ࡣၥ㢟ࡼࡗ࡚␗࡞ࡿ㸬≉㸪ᑐヰᆺࡢ㑇ఏⓗࣝࢦ ࣜࢬ࣒ࡢሙྜࡣ㸪ே㛫ࡢ࣮ࣘࢨࡀほⓗ࡞ホ౯ࢆ⾜࠺ࡇ ࡞ࡿ㸬ࡇࢀࡽࡢฎ⌮ࡢ⧞ࡾ㏉ࡋࡼࡾ㸪ᑡࡋࡎࡘ㞟ᅋ యࡢホ౯್ࡀୖ᪼ࡋ㸪᭱⤊ⓗࡣ᭱Ⰻゎࢆᚓࡽࢀࡿࡇࡀ ᮇᚅࡉࢀࡿ㸬 」ᩘ࣮ࣘࢨࡀཧຍࡋ࡚᥈⣴ࡍࡿࡇ࡛㸪ከࡃࡢ࣮ࣘࢨ ዲࡲࢀࡿゎࢆぢࡘࡅฟࡍᡭἲࡶᥦࡉࢀ࡚࠸ࡿ9-12)㸬୍ ࡋ࡚ᅗ1 ࡢࣇ࣮ࣟࢆ⏝࠸ࢀࡤ㸪A ࡢ⟠ᡤ࠾࠸࡚ಶయࡢホ ౯್ࡀᐃࡲࡗࡓᚋ࡛㸪ಶࠎࡢ࣮ࣘࢨࡢⰋゎࢆࡢ࣮ࣘࢨ ㏦ࡾฟࡍ୍᪉࡛㸪B ࡢ⟠ᡤࡢࡼ࠺᪂ࡓ࡞ୡ௦ࡢಶయ㞟ᅋ ࢆసࡾฟࡍ㝿ࡢ࣮ࣘࢨࡢⰋゎࢆཷࡅྲྀࡾ㞟ᅋࡢ୍ဨ ࡍࡿࡼ࠺࡞ᡭἲࡀ⪃࠼ࡽࢀ㸪ࡇࡢࡼ࠺࡞ᡭἲࡣᓥࣔࢹࣝࢆ ⏝࠸ࡓᡭἲࡋ࡚ᐇ⌧ࡉࢀ࡚࠸ࡿ㸬ࡢ᪉ἲࡋ࡚ࡣ㸪」 ᩘࡢ࣮ࣘࢨࡀ୍⥴ホ౯ࡍࡿࡼ࠺࡞ᡭἲࡶ࠶ࡿ㸬 ࠑ2㺃2ࠒ 㤶ࡾࡢ᭱㐺ゎ᥈⣴ ᑐヰᆺ㐍ィ⟬ࡼࡿ㤶 ࡾࡢ᭱㐺ࡣ㸪ணࡵᐃࡵࡽࢀࡓᗄࡘࡢཎ㤶ᩱࢆࡶ ΰྜࢆ⾜࠺㝿㸪ࡑࢀࡽࡢᙉࡉࢆ᭱㐺ࡍ࡞ࢃࡕ࣮ࣘࢨࡢ ឤᛶࡸዲࡳྜࡗࡓᩘ್ࡋ࡚ᚓࡿࡇࢆᣦࡍ㸬ከᵝ࡛࠶ ࢁ࠺࣮ࣘࢨࡢዲࡳࡸឤᛶ㸪ࡉࡽࡣ⏝┠ⓗᛂࡌࡓ㤶ࡾ ࢆᚓࡿࡇࡀ࡛ࡁࢀࡤ㸪ᵝࠎ࡞ሙ㠃࡛ࡼࡾຠᯝࡢ㧗࠸㤶ࡾ ࢆᚓࡽࢀࡿ࡛࠶ࢁ࠺㸬 ࡇࢀࡽࡢᡭἲ࡛⾜ࡗࡓලయⓗ࡞タィ࡛ࡣ㸪ᅗ 2 ᴫㄝࡍ ࡿࡼ࠺㸪ྛಶయࡢᣢࡘDḟඖࡢኚᩘࡢࡑࢀࡒࢀࡀ㤶ᩱࡢ ᙉࡉᑐᛂࡍࡿ㸬ࡇࢀࡲ࡛㸪ዲࡳྜ࠺㤶ࡾ4,5,13)ࡔࡅ࡛࡞ ࡃ㸪㞟୰࡛ࡁࡿ㤶ࡾ14)㸪ࡉࡽࡣࣜࣛࢵࢡࢫ࡛ࡁࡿ㤶ࡾ15) ࡞ࢆ⏕ᡂࡍࡿ࠸࠺┠ⓗ࡛ࡶ◊✲ࢆ⾜ࡗ࡚ࡁࡓ㸬 ᅗ1 ᑐヰᆺ㑇ఏⓗࣝࢦࣜࢬ࣒ࡢฎ⌮ࣇ࣮ࣟ fig. 1. The flow chart of Interactive GeneticAlgorithm. ୍⯡ⓗ࡞ᑐヰከᆺ㑇ఏⓗࣝࢦࣜࢬ࣒ࡢฎ⌮ࡣ㸪 ಶయ㞟ᅋࡢタᐃ㸪ホ౯㸪㑅ᢥ㸪ཫ✺↛ኚ␗ ࡽᵓᡂࡉࢀࡿ㸬ᅗ୰ࡢࠐࡣ㸪」ᩘ࣮ࣘࢨࡀཧຍࡍ ࡿ㝿㸪㸦A㸧ࡢ࣮ࣘࢨࡢⰋಶయࢆཷࡅྲྀࡿ⟠ᡤ㸪 㸦B㸧⮬㌟ࡢ᥈⣴࡛సࡽࢀࡓⰋಶయࢆ㏦ࡾฟࡍ⟠ᡤ ࢆ♧ࡍ㸬 ᅗ2 ᑐヰᆺ㐍ィ⟬ࡼࡿ࣮ࣘࢨྜ࠺㤶ࡾ᥈⣴ᡭἲࡢ ᴫᛕᅗ
fig. 2. A Schema of Interactive Evolutionary Computation searching a fragrance suited to
user’s preference. ᑐヰᆺ㐍ィ⟬ࡼࡿ㤶ࡾ᥈⣴࡛ࡣ㸪ࢩࢫࢸ࣒ ࡽ㤶ࡾࡀᥦ♧ࡉࢀ㸪ࡑࢀࢆ࣮ࣘࢨࡀホ౯ࡍࡿᙧ࡛ ฎ⌮ࡀ㐍ࡴ㸬ᅗୖ㒊ࡢᩘ್ࡣ㸪࣮ࣘࢨ࠼ࡽࢀ ࡿ㤶ࡾࢆᵓᡂࡍࡿཎ㤶ᩱࡢᙉࡉ࡛࠶ࡾ㸪ࡇࢀࡽࡢ ᩘ್ࡢ⤌ࡳྜࢃࡏࡢ᭱Ⰻゎࡀ᥈⣴ࡢᑐ㇟࡛࠶ࡿ㸬
㐍ィ⟬ࢆ⏝࠸ࡓ」ᩘ࣮ࣘࢨዲࡲࢀࡿ㤶ࡾࡢ᥈⣴ ࠑ2㺃3ࠒ 」ᩘ࣮ࣘࢨࡀཧຍࡍࡿᑐヰᆺ㐍ィ⟬ࡼࡿ㤶 ࡾࡢ᭱㐺ゎ᥈⣴ࢩࢫࢸ࣒ᵓ⠏ ᥦᡭἲࡣ㸪」ᩘ࣮ࣘ ࢨඹ㏻ࡋ࡚ዲࡲࢀࡿ㤶ࡾࢆ᥈⣴ࡍࡿࡇࢆ┠ⓗࡋ࡚࠾ ࡾ㸪ᅗ1 ᅗ 2 ࢆ⤌ࡳྜࢃࡏࡓෆᐜ࡞ࡿ㸬ࡇࢀࡲ࡛㸪⏬ ീࡸ㡢ࢆᑐ㇟ࡋࡓ」ᩘ࣮ࣘࢨࡀཧຍࡍࡿᑐヰᆺ㐍ィ⟬ ࡣᥦࡉࢀ࡚ࡁࡓࡀ㸪㤶ࡾࡘ࠸࡚ࡣ࡞ࡉࢀ࡚࠸࡞࠸㸬ඹ ㏻ࡋ࡚ዲࡲࢀࡿ㤶ࡾࢆぢࡘࡅฟࡍࡇࡣ㸪㤶ࡾࢆ〇ရࡑࡢ ࡶࡢ㸪࠶ࡿ࠸ࡣ〇ရῧຍࡍࡿᙧ࡛ࡢ⏝ྥ࠸࡚࠸ࡿ ⪃࠼ࡽࢀࡿ㸬 ᮏ◊✲࡛ࡣ㸪ᥦᡭἲᇶ࡙ࡁ㸪ᐇ㝿ࢩࢫࢸ࣒ࢆᵓ⠏ ࡋࡓ㸬᥈⣴ࣝࢦࣜࢬ࣒ࡣ㸪㑇ఏⓗࣝࢦࣜࢬ࣒ࢆ⏝࠸ ࡓ㸬2 ྡࡀཧຍࡍࡿᇶ♏ⓗ࡞ࢩࢫࢸ࣒ࡋࡓࡓࡵ㸪ᅗ 1 ࠶ ࡿࡼ࠺࡞ࣇ࣮ࣟ㤶ࡾᥦ♧⨨ࢆ 2 ࢭࢵࢺタࡅࡿࡇ࡞ ࡿ㸬㤶ࡾᥦ♧⨨ࡣ㸪6 ✀ࡢཎ㤶ᩱࢆΰྜྍ⬟࡞࣐ࣟࢪ ࣮ࣗࣝࢆ⏝࠸ࡓ㸬ΰྜ࠾࠸࡚ࡣ㸪0㹼100 ࡢ௵ពࡢ್࡛ྛ ཎ✏ᩱࡢᙉࡉࢆタᐃ࡛ࡁࡿ㸬ࡇࡢ≉ᛶࢆ⏝ࡋ㸪ᑐヰᆺ㑇 ఏⓗࣝࢦࣜࢬ࣒ࡢಶయࡢኚᩘࡣ6 ࡋࡓ㸬ࡲࡓ㸪101 ẁ 㝵ࡢᙉࡉࡢタᐃࡢࡲࡲࡔ㸪ᑠࡉ࡞ᩘ್ࡢኚࡘ࠸࡚ࡣ ࢇ㤶ࡾࡢᙉࡉࡢ㐪࠸ࡢุูࡀࡘ࡞࠸ࡓࡵ㸪ኚᩘࡢ ⠊ᅖࢆ0㹼20 ࡢ 21 ẁ㝵ࡋࡓ㸬ᐇ㝿ࡢ㤶ࡾࡢᥦ♧࠾࠸࡚ ࡣ㸪್ࢆ 5 ಸࡋ࡚࣮ࣘࢨᥦ♧ࡍࡿࡇࡋࡓ㸬ḟ❶࡛⾜ ࠺ᇶ♏ⓗ࡞ᐇ㦂ࡢࡓࡵࡢタᐃࢆ⾲ 1 ♧ࡍ㸬࡞࠾㸪ゎ ࡘ࠸࡚ࡣ㸪」ᩘ࣮ࣘࢨࡼࡿ࣓ࣟࢹ⏕ᡂࡢඛ⾜◊✲12) ࢆཧ⪃㸪ẖୡ௦ࡢಶయ㞟ᅋࡢホ౯ࡀ⤊ࢃࡗࡓⅬ࡛ୡ௦ ୰ࡢ᭱Ⰻゎࢆእ㒊グ᠈⨨㏦ࡾฟࡋ㸪ወᩘୡ௦ࡢ᭱ᚋࡢ ฎ⌮ࡋ࡚እ㒊グ᠈⨨ࡽ┦ᡭࡢ᭱Ⰻゎࢆཷࡅྲྀࡿࡇ ࡋࡓ㸬 ⾲1 ヨసࡋࡓࢩࢫࢸ࣒ࡢタᐃ
Table 1. The parameters of Interactive Genetic Algorithm used in the constructed system.
ཧຍ࣮ࣘࢨᩘ ྡ ୡ௦ᩘ ୡ௦ ಶయᩘ ಶయ 㑅ᢥ ࢚࣮ࣜࢺಖᏑ࠾ࡼࡧࢺ ࣮ࢼ࣓ࣥࢺ㑅ᢥ ཫ Ⅼཫ㸦㸣㸧 ✺↛ኚ␗ s㸦㸣㸧 ᥦᡭἲࡢᇶ♏ⓗ࡞ືస᳨ドࡢࡓࡵᵓ⠏ࡋࡓࢩ ࢫࢸ࣒࠾ࡅࡿᑐヰᆺ㑇ఏⓗࣝࢦࣜࢬ࣒ࡢࣃࣛ ࣓࣮ࢱࢆ♧ࡍ㸬
3. ᐇ㦂
⚟ᒸᕤᴗᏛC21 ᐊ࡚㸪2 ྡࡢ⿕㦂⪅ࡼࡿືస᳨ド ࡢࡓࡵࡢᐇ㦂ࢆ⾜ࡗࡓ㸬2 ྡࡢ⿕㦂⪅ࡣ㸪ᐇ㦂⪅ࡽ᧯స࠾ ࡼࡧホ౯᪉ἲࡢㄝ᫂ࢆཷࡅ㸪⦎⩦ࢆ⾜ࡗࡓୖ࡛㸪ྠ➨0 ୡ௦ࡢホ౯ࢆ㛤ጞࡋࡓ㸬 6 ✀㢮ࡢཎ㤶ᩱࡣ㸪ᑐヰᆺ㐍ィ⟬ࡼࡾዲࡳࡢ㤶ࡾࢆ᥈ ⣴ࡍࡿඛ⾜◊✲13)࡛⏝࠸ࡓ࢜ࣞࣥࢪ㸪࣋ࣝ࢞ࣔࢵࢺ㸪ࣞࣔ ࣥ㸪ࢩࢺࣟࢿࣛ㸪࣒ࣛ㸪ࢢ࣮ࣞࣉࣇ࣮ࣝࢶࡋࡓ㸬ࡇࢀ ࡽࡢ㤶ࡾࡣ㸪࣐ࣟࣞࢩࣆ 1)࡛㢖⦾⏝࠸ࡽࢀࡿ㤶ࡾ࡛࠶ ࡿ㸬࡞࠾㸪ࡇࢀࡽࡢ㤶ࡾᑐࡋ࡚ணࡵ⿕㦂⪅ࡀᣢࡗ࡚࠸ࡿ ༳㇟ࡼࡿホ౯ࢆ㜵ࡄࡓࡵ㸪࠺࠸ࡗࡓཎ㤶ᩱࢆ⏝࠸࡚࠸ ࡿࡘ࠸࡚ࡣ⿕㦂⪅ᩍ♧ࡋ࡞ࡗࡓ㸬ࡲࡓ㸪ࡶ࠺୍᪉ ࡢ⿕㦂⪅ࡽཷࡅྲྀࡗࡓಶయࡀࢀ࡛࠶ࡿࡘ࠸࡚ࡶ㸪 ⿕㦂⪅ࡣᩍ♧ࡋ࡞ࡗࡓ㸬 ⿕㦂⪅ࡣ㸪࣐ࣟࢪ࣮ࣗࣝࢆ㏻ࡌ࡚ࢩࢫࢸ࣒ࡽᥦ♧ࡉ ࢀࡿ㤶ࡾࢆႥࡂ㸪7 ẁ㝵࡛ዲࡳࡢ⛬ᗘࢆホ౯ࡋࡓ㸦1㸸㠀ᖖ ᎘࠸㸪4㸸ࡕࡽ࡛ࡶ࡞࠸㸪㸵㸸㠀ᖖዲࡁ㸧㸬⾲ 1 ♧ ࡋࡓࡼ࠺㸪ୡ௦ᩘ10㸪ಶయᩘ 8 ࡛࠶ࡿࡓࡵ㸪ྛ⿕㦂⪅ࡣ 80 ᅇࡢホ౯ࢆ⾜ࡗࡓ㸬Ⴅぬ⑂ປࢆ㜵ࡄࡓࡵ㸪⿕㦂⪅ࡣ⮬⏤ ఇ᠁ࢆࡿࡇࡀチྍࡉࢀࡓ㸬ࡲࡓ㸪ྠᵝࡢ┠ⓗ࡛㸪⿕ 㦂⪅ᑐࡋ㸪㤶ࡾࡢホ౯ࡢ㛫ࢥ࣮ࣄ࣮ࡢ㤶ࡾࢆႥࡄࡇ 15)ࢆ່ࡵࡓ㸬4. ᐇ㦂⤖ᯝ
ᅗ3㸪4 㸪2 ྡࡢ⿕㦂⪅ A㸪B ࡢࡑࢀࡒࢀࡢホ౯್ࡢ᥎ ⛣ࢆ♧ࡍ㸬ࢢࣛࣇࡣ㸪ୡ௦ࡈࡢホ౯್ࡢᖹᆒ್್᭱ ࡛࠶ࡿ㸬୧⿕㦂⪅ࡶ㸪ୡ௦ࡀ㐍ࡴࡘࢀ࡚ホ౯್ࡀୖ᪼ ࡍࡿഴྥ࠶ࡿࡇࡀࢃࡿ㸬ࡓࡔࡋ㸪⿕㦂⪅B ࡛ࡣୖ᪼ ഴྥࡀᙉࡃᖹᆒ್ࢆぢࡿ⥺ᙧ㏆࠸ࡢᑐࡋ㸪⿕㦂⪅ A ࡛ࡣࡑࡢഴྥࡣᙅࡃ㸪➨3㸪➨ 7 ୡ௦࡞࡛ⱝᖸࡢホ౯್ࡢ పୗࡀほᐹࡉࢀࡓ㸬 ࡲࡓ㸪ࢩࢫࢸ࣒ࡢᇶᮏືసࡋ࡚㸪ẖୡ௦ࡢ⤊ࢃࡾୡ ௦ࡢ᭱Ⰻಶయࡀእ㒊グ᠈⨨᭩ࡁ㎸ࡲࢀࡓࡇ㸪࠾ࡼࡧ㸪 ወᩘୡ௦ࡢ᭱ᚋ┦ᡭࡢ᭱Ⰻಶయࢆཷࡅྲྀࡗࡓࡇࢆ☜ㄆ ࡋࡓ㸬⾲ 2 㸪ྛ⿕㦂⪅ࡢホ౯ࡋ࡚㸪┦ᡭࡽཷࡅྲྀࡗ ࡓಶయࡅࡓホ౯್ࢆ♧ࡍ㸬ᅗ3㸪4 ⾲ 2 ࢆẚ㍑ࡍࡿ㸪 ࡶ࠺୍᪉ࡢ⿕㦂⪅ࡽཷࡅྲྀࡗࡓಶయࡢホ౯್ࡣ㸪㧗ࡃ ࡶୡ௦୰ࡢᖹᆒ್⛬ᗘ࡛࠶ࡾ㸪ᢤࡁࢇ࡛ࡓホ౯್࡛ࡣ↓ ࡗࡓ㸬 ᅗ3 ⿕㦂⪅A ࡢホ౯್ࡢ᥎⛣fig. 3. The progress of subjective fitness value in the subject A.
⿕㦂⪅A ࡢホ౯್ࡢ᥎⛣ࢆ♧ࡍ㸬◚⥺ࡣୡ௦ࡈ ࡢᖹᆒ್㸪ᐇ⥺ࡣ್࡛᭱࠶ࡿ㸬యⓗࡣୖ᪼ ഴྥ࡛࠶ࡿࡀ㸪ᗄࡘࡢୡ௦࡛ホ౯್ࡀୗࡀࡗ࡚ ࠸ࡿࡇࡀࢃࡿ㸬
⚟ᮏㄔ㸪ཎᾏ
ᅗ4 ⿕㦂⪅B ࡢホ౯್ࡢ᥎⛣
fig. 4. The progress of subjective fitness value in the subject B.
⿕㦂⪅B ࡢホ౯್ࡢ᥎⛣ࢆ♧ࡍ㸬◚⥺ࡣୡ௦ࡈ ࡢᖹᆒ್㸪ᐇ⥺ࡣ್࡛᭱࠶ࡿ㸬ᖹᆒ್㸪್᭱ ࡀࡶୖ᪼ഴྥ࠶ࡿࡇࡀࢃࡿ㸬
⾲2 ┦ᡭࡽཷࡅྲྀࡗࡓಶయࡢホ౯್ Table 2. The fitness values on the individuals
which was transmitted by another subject.
ୡ௦ 2 4 6 8 ⿕㦂⪅A 2 2 4 3 ⿕㦂⪅B 4 2 4 3 እ㒊グ᠈⨨ࢆ㏻ࡌ࡚ࡶ࠺୍᪉ࡢ⿕㦂⪅ࡽཷࡅ ྲྀࡗࡓಶయᑐࡋ࡚㸪ࡑࢀࡒࢀࡢ⿕㦂⪅ࡀࡅࡓ ホ౯್ࢆࡲࡵࡓ⤖ᯝࢆ♧ࡍ㸬ወᩘୡ௦ࡢ᭱ᚋ ཷࡅྲྀࡾḟୡ௦ࡢ㞟ᅋྵࡵࡿࡓࡵ㸪അᩘୡ௦ ホ౯ࡍࡿࡇ࡞ࡿ㸬
5. ⪃ᐹ
ᥦᡭἲࢆලయⓗ࡞ࢩࢫࢸ࣒ࡋ࡚ᵓ⠏ࡋ㸪2 ྡࡢ⿕㦂⪅ ࡢࡳ࡛ࡣ࠶ࡿࡀᐇ㦂ࢆ⾜࠸㸪ᇶᮏⓗ࡞ືసࡢ᳨ドࢆ⾜ࡗࡓ㸬 ᐇ㦂⤖ᯝࡋ࡚㸪ྛ⿕㦂⪅ࡢᐇ㦂ࡢ㐣⛬࡛ᚓࡽࢀࡓಶయ ࡑࢀᑐࡍࡿホ౯್ࢆほᐹࡍࡿࡇࡼࡾ㸪እ㒊グ᠈⨨ ࢆ㏻ࡌࡓᇶᮏⓗ࡞Ⰻゎࡢࡀ㸪ィ⏬㏻ࡾືసࡋ࡚࠸ࡿ ࡇࢆ☜ㄆࡋࡓ㸬ࡇࡢࢩࢫࢸ࣒࡛ࡣ㸪ྛ࣮ࣘࢨࡢⰋゎࢆእ 㒊グ᠈⨨㏦ࡾ㸪ࡲࡓཷࡅྲྀࡿ࠸࠺༢⣧࡞ฎ⌮࡛ゎ ࢆ⾜ࡗ࡚࠸ࡿ㸬ࡑࡢࡓࡵ㸪ཧຍ⪅ᩘࢆቑࡸࡋ࡚ࡶ㸪ྛࣘ ࣮ࢨࡽࡢእ㒊グ᠈⨨ࡢࢡࢭࢫࡉ࠼࡛ࡁࢀࡤ㸪༠ㄪ సᴗࡢࡼ࠺࡞ᙧ࡛ゎࡢ᥈⣴ࡀྍ⬟࡛࠶ࡿ㸬ࡲࡓ㸪ゎࡢ ࢱ࣑ࣥࢢࡋ࡚ࡣ㠀ྠᮇ㸪ࡍ࡞ࢃࡕ࣮ࣘࢨ㛫࡛ୡ௦᭦᪂ ࡢࢱ࣑ࣥࢢ࡞ࢆᥞ࠼ࡿᚲせࡢ࡞࠸ࢩࢫࢸ࣒࡞ࡗ࡚࠸ ࡿ㸬ࡇࡢ≉ᛶࡽ㸪ከࡃࡢ࣮ࣘࢨࡀ⮬⏤ཧຍྍ⬟࡞ࢩࢫ ࢸ࣒ゝ࠼ࡿ㸬 ࡲࡓ㸪ホ౯್ࡢ᥎⛣ࢆୡ௦ࡈࡢᖹᆒ್್᭱ࡽほ ᐹࡋࡓ㸬2 ྡࡢ⿕㦂⪅ࡢࡳࡢࡓࡵ㸪⤫ィⓗ࡞ゎᯒ࡞ࡣ࡛ࡁ ࡞࠸ẁ㝵࡛࠶ࡿࡀ㸪ᴫࡡⰋዲ࡞⤖ᯝゝ࠼ࡿ㸬ࡓࡔࡋ㸪⿕ 㦂⪅A ࡢ⤖ᯝࢆぢࡿ㸪༢ㄪ࡞ୖ᪼ࡀぢࡽࢀࡿ࠸࠺ࢃࡅ ࡛ࡣ࡞࠸ࡼ࠺࡛࠶ࡿ㸬࢚࣮ࣜࢺಖᏑࢆ⏝࠸࡚࠸࡚ࡶࡇ࠺࠸ ࡗࡓ᥎⛣ࡀ㉳ࡁ࠺ࡿࡇࡣᑐヰᆺ㐍ィ⟬ࡢ≉ᛶ࡛ࡣ࠶ࡿ ࡶࡢࡢ㸪ཎᅉࡢㄪᰝࡀᚲせᛮࢃࢀࡿ㸬」ᩘࡢ࣮ࣘࢨࡼ ࡾ࣓ࣟࢹࢆ⏕ᡂࡋࡓඛ⾜◊✲12)࡛ࡣ㸪ࡢ࣮ࣘࢨࡽཷ ࡅྲྀࡗࡓಶయࡢホ౯್ࢆ᳨ドࡍࡿࡇ࡛㸪ᡭἲࡢ᭷ຠᛶࢆ ㄪᰝࡋࡓ㸬ᮏ◊✲࡛ᥦࡋࡓᡭἲࡘ࠸࡚ࡶ㸪ከࡃࡢ⿕㦂 ⪅ࡼࡿᐇ㦂ࢆ⾜ࡗࡓᚋ㸪ྠᵝࡢゎᯒࢆ⾜࠸㸪᭷ຠᛶࡢ᳨ ドࢆ⾜࠺ᚲせࡀ࠶ࡿ㸬 ゎࡢຠᯝࢆㄪᰝࡍࡿࡓࡵ㸪ࡇࢀࡽࡢಶయࡢホ౯ࡸ ゎࡢຠᯝࢆㄪᰝࡍࡿࡇࡣ㔜せ࡞ㄢ㢟࡛࠶ࡿ㸬⾲ 2 ♧ࡋࡓࡼ࠺㸪┦ᡭࡽ㏦ࡽࢀ࡚ࡁࡓಶయࡢホ౯್ࡣ㸪 㧗ࡃࡶୡ௦୰ࡢᖹᆒ್⛬ᗘ࡛࠶ࡾ㸪ゎࡢᙉ࠸ຠᯝࡀ ࠶ࡿࡣゝ࠸㞴࠸㸬⿕㦂⪅㛫࡛ఝࡓࡼ࠺࡞ឤᛶࢆᣢࡗ࡚࠸ ࡚ࡶ㸪᥈⣴ึᮇಶయ㞟ᅋࡢᅇ✵㛫୰ࡢ⨨ࡀࡁࡃ␗࡞ ࡿࡇࡶ࠶ࡾ㸪ࢹ࣮ࢱࢆ㞟ࡵ㸪ࡼࡾ㛗࠸ୡ௦ᩘࡢᐇ㦂ࢆ⾜ ࡗࡓᚋ㸪ࡇࡢࡼ࠺࡞㆟ㄽࢆ⾜࠺ணᐃ࡛࠶ࡿ㸬࡞࠾㸪ᅇ ࡢᐇ㦂࡛ࡣ㸪ࡶ࠺୍᪉ࡢ⿕㦂⪅ࡽཷࡅྲྀࡗࡓಶయࡀࢀ ࡛࠶ࡿࡘ࠸࡚ࡣᩍ♧ࢆ⾜ࢃ࡞ࡗࡓࡀ㸪ゎࢆ⾜ࡗ ࡓඛ⾜◊✲16)࡛ࡣ㸪ᙜヱಶయࡀࢀ࡛࠶ࡿࢆ▱ࡿࡇࡀ ༠ㄪసᴗࡢຠ⋡ࢆୖࡆࡿࡢ࡛ࡣ࡞࠸㸪㏙࡚࠸ࡿ㸬ࡇ ࡢࡼ࠺࡞ほⅬࡽ㸪ᩒ࠼࡚ࡇࡢᩍ♧ࢆ⾜࠺ࡇࡘ࠸࡚ࡶ ᳨ウࢆ㐍ࡵࡓ࠸㸬ᅇࡣ 2 ྡࡢࡳࡢ⿕㦂⪅࡛࠶ࡗࡓࡓࡵ㸪 ࡇࡢゎᯒࡘ࠸࡚ࡶ㸪ࡼࡾከࡃࡢ࣌ࡼࡿᐇ㦂ࡀᚲせ࡛ ࠶ࡿ㸬6. ⤖ゝ
ᮏ◊✲࡛ࡣ㸪」ᩘ࣮ࣘࢨࡀཧຍࡍࡿ㤶ࡾ᭱㐺ゎ᥈⣴ࡢࡓ ࡵࡢᑐヰᆺ㐍ィ⟬ࢆᥦࡋࡓ㸬2 ྡࡀཧຍࡍࡿࢩࢫࢸ࣒ࢆ ᵓ⠏ࡋ㸪ゎࡢᶵ⬟ࡀṇᖖືసࡍࡿࡇ㸪⡆༢࡞ᐇ㦂 ࢆ㏻ࡌ࡚ホ౯್ࡢ᥎⛣࡞ࢆㄪᰝࡋࡓ㸬 ㏻ᖖࡢᑐヰᆺ㐍ィ⟬࡛ࡣ㸪ಶࠎࡢ࣮ࣘࢨࡢዲࡳ࣭ឤᛶ ྜ࠺ゎࢆ᥈ࡍࡢᑐࡋ㸪」ᩘ࣮ࣘࢨࡢඹ㏻ࡋࡓዲࡳ࣭ឤ ᛶྜ࠺ゎࢆ᥈ࡋฟࡍⅬ≉ᚩࡀ࠶ࡿࡓࡵ㸪㤶Ỉ࡞ࡢ㤶 ࡾࡑࡢࡶࡢࡢ〇ရࡔࡅ࡛࡞ࡃ㸪㤶ࡾࢆῧຍࡋࡓᵝࠎ࡞〇ရ ࡢ㛤Ⓨࡸࡑࡢ⿵ຓ࡞ࡾ࠺ࡿ⪃࠼࡚࠸ࡿ㸬ᚋࡣ㸪ᡭἲ ࡢ᭷ຠᛶࡢ᳨ドࡔࡅ࡛࡞ࡃ㸪Ⴅぬ⑂ປࡢపῶࡸࣥࢱࣇ࢙ ࣮ࢫࡢᨵⰋࢆ⤒࡚㸪ᐇ⏝ࢆ┠ᣦࡋࡓ࠸㸬ㅰ㎡
ᮏ◊✲ࡣ㸪⚟ᒸᕤᴗᏛ⥲ྜ◊✲ᶵᵓᖹᡂ30 ᖺᗘⱝᡭᩍ ဨ◊✲㧗ᗘᨭไᗘࡼࡿ⿵ຓࢆཷࡅ࡚⾜ࢃࢀࡓ㸬ࡇࡇ ㅰពࢆグࡍ㸬 㸦㸰㸮㸯㸷ᖺ㸯㸮᭶㸯㸶᪥ཷ㸧ᩥ
⊩
(㸯) 㟷ᮌᜨ㸸࣐ࣟ㤶Ỉࡢᡭసࡾࣂࣈࣝ㸪ゅᕝ࣐࢞ࢪࣥࢬ (2008). (㸰) R. Dawkins : “The Blind Watchmaker”, Longman Scientific &Technical (1986).
(㸱) H. Takagi : “Interactive Evolutionary Computation: Fusion of the Capabilities of EC Optimization and Human Evaluation”, Proc.
㐍ィ⟬ࢆ⏝࠸ࡓ」ᩘ࣮ࣘࢨዲࡲࢀࡿ㤶ࡾࡢ᥈⣴ the IEEE, Vol.89, No.9, pp.1275-1296 (2001).
(㸲) M. Fukumoto and J. Imai : “Design of Scents Suited with User’s Kansei using Interactive Evolutionary Computation”, Proc. KEER2010, pp.1016-1022 (2010).
(㸳) M. Fukumoto, M. Inoue, and J. Imai : “User’s Favorite Scent Design Using Paired Comparison-based Interactive Differential Evolution”, Proc. IEEE CEC2010, pp.4519-4524 (2010).
(㸴) M. Fukumoto, K. Kawai, M. Inoue, and J. Imai : “Interactive Tabu Search with Paired Comparison for Optimizing Fragrance”, Proc. IEEE SMC2013, pp.1690-1695 (2013).
(㸵) H. Takagi and D. Pallez : “Paired Comparison-based Interactive Differential Evolution”, Proc. World Congress on Nature and Biologically Inspired Computing, pp.375-380 (2009).
(㸶) M. Fukumoto, S. Koga, M. Inoue, and J. Imai : “Interactive Differential Evolution Using Time Information Required for User’s Selection: In A Case of Optimizing Fragrance Composition”, Proc. 2015 IEEE CEC, pp.2192-2198 (2015). (㸷) Y. Ogawa, M. Miki, T. Hiroyasu, and Y. Nagaya : “A New
Collaborative Design Method Based on Interactive Genetic Algorithms”, Proc. the EUROGEN2001, pp.109-114 (2001). (㸯㸮) H. Takenouchi, H. Inoue, and M. Tokumaru : “Signboard
design system through social voting technique”, Proc. ISIC2014, pp.14-19 (2014).
(㸯㸯) M. Fukumoto and T. Hatanaka : “A Proposal for Distributed Interactive Genetic Algorithm for Composition of Musical Melody”, IEE, Vol.3, No.2, pp.56-68 (2017).
(㸯㸰) K. Nomura and M. Fukumoto : “Music Melodies Suited to Multiple Users’ Feelings Composed by Asynchronous Distributed Interactive Genetic Algorithm”, International Journal of Software Innovation, Vol.6, No.2, pp.26-36 (2018).
(㸯㸱) Ἑྜၨ㸪㡰୍㸪ୖㄔ㸪⚟ᮏㄔ㸸㏆ഐ᥈⣴⠊ᅖࢆῶࡉ ࡏࡿᑐヰᆺࢱࣈ࣮ࢧ࣮ࢳࡼࡿ㤶ࡾ⏕ᡂᡭἲ㸪⏕ࢯࣇࢺ࢙࢘ࢩ ࣏ࣥࢪ࣒࢘2012 ㅮ₇ㄽᩥ㞟㸪G4-3 (2012). (㸯㸲) ⸨㔝ᑗః㸪ྂ㈡ៅᖹ㸪⚟ᮏㄔ㸸ᑐヰᆺࢱࣈ࣮ࢧ࣮ࢳࡼࡿ㞟୰ ࡛ࡁࡿ㤶ࡾࡢ᥈⣴㸸ึᮇಶయ࣐ࣟࣞࢩࣆࢆ⏝࠸ࡓሙྜ㸪⏕ࢯ ࣇࢺ࢙࢘ࢩ࣏ࣥࢪ࣒࢘2014 ㅮ₇ㄽᩥ㞟, G1-2 (2014). (㸯㸳) ᮧభ㸪⚟ᮏㄔ㸸ࢥ࣮ࣄ࣮ࡢ㤶ࡾࢆ⏝࠸ࡓຠ⋡ࡢⰋ࠸Ⴅぬ㡰 ᛂ⦆᪉ἲࡢ᳨ウ㸪⏕ࢯࣇࢺ࢙࢘ࢩ࣏ࣥࢪ࣒࢘2011 ㅮ₇ㄽᩥ 㞟㸪pp.119-121 (2011). (㸯㸴) ᯇᮏᾴᖹ, ୖᮧ᱈Ꮚ, すᆂ, Ώ㑓┿ஓ㸸ேࢤ࣮࣒ᙧᘧࡢ㐍 ⓗ༠ㄪ᭱㐺㸪➨12 ᅇ㐍ィ⟬◊✲㸪P1-12 (2017).
