NLP若手の会 (YANS) 第17回シンポジウム (2022年8月29日)のチュートリアル講演資料です。 雑談などのスライドは省いてあります。
ֶज़ใݕࡧͱਪનܡҪ ຑཬҥಉࢤࣾେֶ ཧֶ෦ ΠϯςϦδΣϯτใֶՊ:"/4
View Slide
ߨԋऀհܡҪ ຑཬҥ݄ ւಓେֶֶ෦ใΤϨΫτϩχΫεֶՊଔۀ݄ ւಓେֶେֶӃใՊֶݚڀՊϝσΟΞωοτϫʔΫઐ߈म࢜՝ఔम݄ྃ ຊֶज़ৼڵձಛผݚڀһ %$݄ ւಓେֶେֶӃใՊֶݚڀՊϝσΟΞωοτϫʔΫઐ߈ത࢜ޙظ՝ఔमྃത࢜ʢใՊֶʣ݄ ຊֶज़ৼڵձಛผݚڀһ 1%݄ ಉࢤࣾେֶཧֶ෦ใγεςϜσβΠϯֶՊॿڭ݄ ಉࢤࣾେֶཧֶ෦ΠϯςϦδΣϯτใֶՊॿڭ݄ ʏ ।ڭतΧλΧφֶՊ͔ΓͷܦྺͰ͢ʜ
ֶज़ݚڀΛऔΓר͘ࣄిࢠδϟʔφϧͷ͋ΔௐࠪʹΑΔͱίϯϐϡʔλαΠΤϯεܥͷݚڀऀ͕ಡΜͰ͍Δจฏۉ ݄ຊʢʹຊʣʂ <5FOPQJS >⻘線:英語論⽂誌の数⾚線:増加率ൈਮݩʀ3+PIOTPO "8BULJOTPO BOE.8BCF l5IF45.SFQPSU"OPWFSWJFXPGTDJFOUJGJDBOETDIPMBSMZQVCMJTIJOH zUIFEJUJPO 0DU$5FOPQJS -$ISJTUJBO +,BVGNBO l4FFLJOH 3FBEJOH BOE6TFPG4DIPMBSMZ"SUJDMFT"O*OUFSOBUJPOBM4UVEZPG1FSDFQUJPOTBOE#FIBWJPSPG3FTFBSDIFST z 1VCMJDBUJPOT WPM OP
ݚڀऀΛࢧ͑ΔݕࡧΤϯδϯIUUQTTDIPMBSHPPHMFDPKQ
(PPHMF4DIPMBSͷݕࡧΞϧΰϦζϜ• ʮจͷϑϧςΩετɺग़൛ॴɺஶऀɺඃҾ༻ճɺඃҾ༻ͷΛߟྀʯCZ(PPHMFIUUQTTDIPMBSHPPHMFDPNJOUMKBTDIPMBSBCPVUIUNM
(PPHMF4DIPMBSͷݕࡧΞϧΰϦζϜ• ʮจͷϑϧςΩετɺग़൛ॴɺஶऀɺඃҾ༻ճɺඃҾ༻ͷΛߟྀʯCZ(PPHMF༗໊ͳஶऀʁ͍ͭͷҾ༻͕ଟ͍ʁ༗໊ͳδϟʔφϧʁඃҾ༻ଟ͍ʁ
ஶऀͷධՁࢦඪ• ࠷؆ศͳࢦඪʮจʯɻจ• ࣭Λߟྀ͍ͯ͠ͳ͍ͱ͍͏͕͋Δɻ
ඃҾ༻• ଞऀ͔Βࢀߟʹ͞ΕΔจ࣭͕ߴ͍ͱԾఆɻࢀߟจݙ
ஶऀͷධՁࢦඪ IJOEFY<)JSTDI 1/"4 >• จͷੜ࢈ੑͱӨڹྗͷ྆ํΛଌఆ͠Α͏ͱ͢ΔࢦඪɻจΠϯσοΫε ඃҾ༻ ඃҾ༻ɿIJOEFYJJOEFY(被引⽤数10以上の論⽂数)"͞Μ+&)JSTDI "OJOEFYUPRVBOUJGZBOJOEJWJEVBMTTDJFOUJGJDSFTFBSDIPVUQVU z 1SPDFFEJOHTPGUIF/BUJPOBM"DBEFNZPG4DJFODF WPM OP QQr • ۙͷؒʹग़൛͞Εͨจʹ͍ͭͯࢉग़ͨ͠߹ɿIJOEFY• Լهͷར͔ΒٸʹΛूΊͨ ɿ• Ҿ༻ͷͳ͍จฏۉҎ্ͷҾ༻ΛͭจͳͲɺۃͳʹӨڹ͞Εͳ͍ɻ• 8FCPG4DJFODFͳͲͷσʔλϕʔεʹऩ͞Ε͍ͯΔඃҾ༻ճΛιʔτ͢Δ͚ͩͰ؆୯ʹࢉग़Ͱ͖Δɻ• ୯ҰͷͷൺֱͰࡁΉɻ
IJOEFYͷੜɺHJOEFY<&HHIF *44*/FXTMFUUFS >l ඃҾ༻ॱͰจΛฒͨͱ͖ɼ্Ґ(ݸͷจ͕গͳ͘ͱ(ճͷҾ༻Λड͚ΔΑ͏ͳʢ།Ұͷʣ࠷େͷɻ• HJOEFY্Ґͷจ͕߹ΘͤͯͷඃҾ༻Λಘͨ͜ͱΛࣔ͢ɻ• ඃҾ༻͕ඇৗʹଟ͍จΛߴ͘ධՁ͢Δ͕IJOEFYͱͷࠩҟɻ• શͯͷൃදจͷฏۉඃҾ༻͕ൃදจͷ૯Λ͑ͨ߹ʹ͏·͘ධՁͰ͖ͳ͍ͱ͍͏ܽΛͭɻl ଞʹIJOEFYͷੜ͕ଟ͘ఏҊ͞Ε͍ͯΔ͕ɺ͍ͣΕࢉग़͕ෳࡶͰ͋ͬͨΓܽΛ๊͍͑ͯΔͨΊɺIJOEFYɺHJOEFYɺ૯ඃҾ༻ͷछྨ͕༻͍ΒΕΔ͜ͱ͕ଟ͍ɻจΠϯσοΫε ඃҾ༻ ඃҾ༻ɿIJOEFYHJOEFY"͞Μ
IJOEFYʹର͢Δʑͷࢦఠ• IJOEFYͷߴ͍ݚڀऀɼଟ͘ͷڞஶऀʢ૯ਓҎ্ʣΛͪɺޓ͍ʹҾ༻͋͠͏͜ͱ͕ଟ͍<.BTJD#FHJD >ɻ• ڞஶจ͕ଟ͍΄ͲจɺඃҾ༻͕૿Ճ͢Δ͜ͱΑ͘ΒΕ͍ͯΔɻ• ஶऀͷIJOEFY͕ʙͰ͋Γɺ΄ͱΜͲͷஶऀޓ͍ʹ۠ผΛ͚ͭΔ͜ͱ͕͍͠<$SFTQP4JNPFT 442 >ɻ• IJOEFYͷࢉग़ରͰ͋ΔIʹೖΒͳ͍จΛධՁͰ͖ͳ͍ɻڞஶऀͷྠඞͣ͠ສೳͳࢦඪͰͳ͍͜ͱʹҙ͢Δඞཁ͕͋Δɻ
ग़൛ॴʢδϟʔφϧɺࠃࡍձٞͳͲʣͷධՁࢦඪ• ஶऀͷͱ͖ͱಉ༷ʹIJOEFYܥͷࢦඪ͕ࢉग़͞ΕΔɻδϟʔφϧIF(y) =ʢy-1ͷܝࡌจʣʴʢy-2ͷܝࡌจʣy-1ɺy-2ͷܝࡌจ͕ZʹҾ༻͞ΕͨճۙؒͷඃҾ༻͕ଟ͍ॱʹจΛιʔτ…IJOEFY• *NQBDUGBDUPS *'ͱ͍͏ࢦඪ༗໊ɻ͜Εग़൛ॴಛ༗ͷࢦඪɻ
Ҿ༻จԽͷҧ͍ʹҙ͢Δඞཁ͕͋Δ• Ҿ༻ͷ͞ΕํʹΑͬͯେ͖͘ҟͳΔͨΊɺಉҰͰͷධՁʹ͔͑͠ͳ͍ɻ• ݄ݱࡏɺ֤ͷIJOEFYτοϓࡶࢽҎԼͷͱ͓Γɻ˞(PPHMF4DIPMBS.FUSJDTΑΓ• ίϯϐϡʔλϏδϣϯɿ$713 *$$7 &$$7 • ܭࢉݴޠֶɿ"$- &./-1 )-5/""$- • ਤॻؗใֶɿ4DJFOUPNFUSJDT +PVSOBMPGUIF"TTPDJBUJPOGPS*OGPSNBUJPO4DJFODFBOE5FDIOPMPHZ +PVSOBMPG*OGPSNFUSJDT • ҨֶɾήϊϛΫεɿ/BUVSF(FOFUJDT (FOPNF#JPMPHZ • ݐஙɿ#VJMEJOHT -JHIUJOH3FTFBSDI5FDIOPMPHZ • Ҿ༻ʹΑͬͯҟͳΔɻ• ࡉ๔ੜֶͷ্Ґࡶࢽͷ*'ɻ• ֶͷ্Ґࡶࢽͷ*'ɻ
(PPHMF4DIPMBSͷΞϧΰϦζϜௐࠪ <#FFM +41 >• ༷ʑͳݕࡧΫΤϦΛ༻͍ͯจݕࡧ݁ՌΛऩू͠ɺඃҾ༻ճɾޠͷग़ݱසɾग़൛ͱදࣔॱҐͷؔΛϓϩοτɻҎԼͷ͜ͱ͕ࣔࠦ͞Εͨɻü ඃҾ༻ճͷଟ͞ॱҐʹେ͖͘Өڹ͢Δɻü ΫΤϦͷޠ͕۟ग़ݱ͢ΔճӨڹ͠ͳ͍ɻͨͩ͠ɺλΠτϧʹೖ͍ͬͯΔํ্͕Ґʹ͘Δɻü ݕࡧΫΤϦͷಉٛޠߟྀ͞Ε͍ͯͳ͍ɻü ΫΤϦޠ۟Λஶऀ໊ʹͭͱ্Ґʹ͘Δɻ+#FFM #(JQQ BOE&8JMEF l"DBEFNJD4FBSDI&OHJOF0QUJNJ[BUJPO "4&00QUJNJ[JOH4DIPMBSMZ-JUFSBUVSFGPS(PPHMF4DIPMBSBOE$P z+PVSOBMPG4DIPMBSMZ1VCMJTIJOH WPM OP QQ +BOVBSZ8IJUF-&%ˠ8IJUF͞Μ͕ݕࡧ݁ՌʹʜඃҾ༻είΞݕࡧ݁ՌͷॱҐಉٛޠΛొͤͨ͞ํ͕ݕࡧʹ༗རʜλΠτϧॏཁʜ
(PPHMF4DIPMBSͷϦόʔεΤϯδχΞϦϯά <3PWJSB 1%-* >• (PPHMF4DIPMBSͷ֤छࢦఆݕࡧʹΑΔॱҐඃҾ༻͚ͩͰ࠶ݱͰ͖Δ͔Λௐࠪɻ• Ͳͷݕࡧํ๏ʹ͓͍ͯඃҾ༻ʹΑΔϥϯΩϯά݁ՌͱͷεϐΞϚϯॱҐ૬͕ؔҎ্ͱͳͬͨɻ໊ࣈͰݕࡧग़൛ॴ໊ͰݕࡧͰݕࡧඃҾ༻ʹΑΔϥϯΩϯά$3PWJSB '(VFSSFSP4PMÉ -$PEJOB l3FDFJWFEDJUBUJPOTBTBNBJO4&0 GBDUPSPGHPPHMFTDIPMBSSFTVMUTSBOLJOH z1SPGFTJPOBMEF-B*OGPSNBDJPO WPM OP QQr ඃҾ༻͕ଟ͍จ͕ࠓޙΛूΊΔͱ͍͏ϚγϡʔޮՌΛ૿େͤ͞ΔΈʁ
༨ஊɿ(PPHMF4DIPMBS͔Βͷσʔλऩू• σʔλऩू༻"1*ఏڙ͞Ε͍ͯͳ͍ʢʂʣ• εΫϨΠϐϯάΛ܁Γฦ͢ʹ$"15$)"Λղ͘ඞཁ͕͋Δɻ• (PPHMF4DIPMBS্ͷສຊͷจσʔλΛੳͨ͠νʔϜɺϲ݄͔͚ͯखಈͰ$"15$)"Λղ͖ଓ͚ͨͱͷ͜ͱ <&MTF>ɻ)&MTF l)PX*TDSBQFEEBUBGSPN(PPHMF4DIPMBS z/BUVSF/&842" IUUQTXXXOBUVSFDPNBSUJDMFTEϙνʜϙνʜ
จݕࡧͱจਪન• จΛݕࡧ͢Δʹॴͷ༰ΛΫΤϦͱͯ͠໌֬ʹೖྗ͢Δඞཁ͕͋Δɻ• Ұํɺؔ࿈͢ΔจΛޮతʹΓ͍ͨͱ͍͏χʔζʹ͑Δͷ͕จਪનͰ͋Δɻ(MPCBMͳจਪન MPDBMͳจਪનೳಈతʹݕࡧ͢Δͷ͕ͭΒ͍ͷʜؔ࿈͢ΔจΛ͓͢͢Ί͢ΔΑʂࢲʹͽͬͨΓͷจͩΘʂʂʂจষʹର͠Ҿ༻͖͢จݙΛਪનl-PDBUJOHTDJFOUJGJDTQFDJGJDDPOUFOUGSPNBMBSHFDPSQPSBJTDSVDJBMUPSFTFBSDIFST< >zจͷओͳ༰ʹຊߨԋͰͪ͜ΒΛհ
จਪનख๏ͷΧςΰϥΠζڠௐϑΟϧλϦϯάख๏ ίϯςϯτϕʔεख๏άϥϑຒΊࠐΈख๏ ϋΠϒϦουϢʔβͷڵຯ จͷ༰4JNJMBS4JNJMBSҾ༻ Ҿ༻
ڠௐϑΟϧλϦϯά DPMMBCPSBUJWFGJMUFSJOH$'• ଞϢʔβͷධՁใΛ༻͍Δਪનख๏ɻ• &ίϚʔεͰͷਪનͱҟͳΓɺจਪનͰϢʔβΛߟ͑ͣࢀߟจݙϦετΛ༻͍Δ͜ͱ͕ଟ͍ɻØ จʹϨʔςΟϯάΛ༩͑Δͷ໘ͳͨΊ <:BOH+$%->ɻ映画1 映画2 映画3 映画4 …Ϣʔβ5 3 2Ϣʔβ3 2 3Ϣʔβ5 2 3ʜ論⽂1 論⽂2 論⽂3 論⽂4 …จ0 1 0 0จ0 0 0 0จ0 0 0 1ʜөըͷਪન ߦ͕Ϣʔβɺྻ͕ΞΠςϜͷϨʔςΟϯάߦྻจਪન ߦ͕Ҿ༻ݩจɺྻ͕Ҿ༻ઌจͷೋߦྻ
ڠௐϑΟϧλϦϯάʹΑΔจਪન <.D/FF$4$8`>• ߦͱྻΛจͱͯ͠ߏஙͨ͠Ҿ༻ߦྻʹ͓͍ͯ L/FBSFTU/FJHICPS L//$'Λ࣮ߦɻ• ߦϕΫτϧؒͷίαΠϯྨࣅΛࢉग़͠ɺΫΤϦจͱͷྨࣅ্͕Ґ L ݅ͱͳΔจΛબग़ɻ• ֤ۙͷҾ༻ঢ়گΛྨࣅͰॏΈ͚ͯ͠ථɻ… p50p51p52p53p54 …p1 p2 p3 p4 p5 …………………ΫΤϦจTJNJMBSLۙL͔ۙΒͷථ ֤ۙͱͷྨࣅ4..D/FF FUBM l0OUIFSFDPNNFOEJOHPGDJUBUJPOTGPSSFTFBSDIQBQFST zJO1SPD$4$8 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Pa,i = ¯ra +Pu2N(ru,i ¯ru) ⇥ sim(a, u)Pu2Nsim(a, u)˞Ұൠʹ6TFSCBTFEL//$'ͱݺΕΔࣜpa, piؒͷྨࣅࢀߟจݙ
ڠௐϑΟϧλϦϯάʹΑΔจਪન <.D/FF$4$8`>• ྻϕΫτϧؒʢʹඃҾ༻จؒʣͷྨࣅΛར༻͢Δόʔδϣϯ͋Δɻ… p50p51p52p53p54 …p1 p2 p3 p4 p5 …………………ΫΤϦจ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Pa,i =Pj2Nira,j⇥ sim(i, j)Pj2Nisim(i, j)˞Ұൠʹ*UFNCBTFEL//$'ͱݺΕΔࣜLۙL͔ۙΒͷථ ֤ۙͱͷྨࣅࢀߟจݙ• ͍ͣΕͷ$'(PPHMFݕࡧʹൺͯ৽ͨͳจΛൃݟ͢Δͷʹཱͬͨɻ
ڠௐϑΟϧλϦϯάͰ༻͍Δߦྻͷվྑ <-JV*&&&"DDFTT >• ଞͷҾ༻ݩจݙ͔ΒજࡏతͳؔΛਪఆ͢Δ$POUFYUCBTFE$PMMBCPSBUJWF'JMUFSJOHΛఏҊɻ)-JV 9,POH 9#BJ 88BOH 5.#FLFMFBOE'9JB $POUFYU#BTFE$PMMBCPSBUJWF'JMUFSJOHGPS$JUBUJPO3FDPNNFOEBUJPO *&&&"DDFTT WPM QQ p5p6p1p2p3p4p1ͱ p4ಉ͡จΛҾ༻͍ͯ͠ͳ͍͕ɺp2 p3͔Βؔ࿈ੑΛݟ͍ͩͤΔp2͔ΒҾ༻ p2͔ΒҾ༻͞Ε͍ͯͳ͍p1͔ΒҾ༻ N11N12p1͔ΒҾ༻͞Ε͍ͯͳ͍N21N22ˠׂද͔Β p1ͱ p2ͷؔͷڧ͞ΛఆྔԽ͠ɺᮢॲཧͯ͠$'༻ͷߦྻཁૉͱ͢Δɻ• ,%%$VQσʔληοτͰ࣮ݧͨ݁͠Ռɺී௨ͷ$'ΑΓߴਫ਼ͳਪનΛୡͨ͠ɻ
ڠௐϑΟϧλϦϯάϕʔεͷ• ߦྻ͕ඇৗʹεύʔεͰ͋Γɺ͔ͭཁૉͷ༏ྼ͕ෆ໌ྎɻ映画1 映画2 映画3 映画4 …Ϣʔβ5 3 2Ϣʔβ3 2 3Ϣʔβ5 2 3ʜ論⽂1 論⽂2 論⽂3 論⽂4 …จ0 1 0 0จ0 0 0 0จ0 0 0 1ʜөըͷਪન ü Ϣʔβ͕ͨ͘͞Μ͍Δü ͖ݏ͍ͷใ͕͋Δจਪન ü Ұͭͷจ͔Βͷථ͕গͳ͍ü ೋͰ͋ΓҾ༻Λશͯ͘͠ѻ͏
ίϯςϯτϕʔεͳจਪન• จͷλΠτϧɺΞϒετϥΫτɺΩʔϫʔυͳͲͷςΩετใ͔Βஶऀจͷ༰Λਪఆɻ• ύʔιφϥΠθʔγϣϯਪનͳͲΑ͘ఏҊ͞Ε͍ͯΔɻҙ• ઐʹಛԽ͗͢͠Δͱɺ͢Ͱʹ͍ͬͯΔจ͔Γ͕ఏҊ͞ΕΔͱ͍͏͕͋Δɻ• จͷ࣭ਓؾແࢹ͞ΕΔ͜ͱ͕ଟ͍ɻจਪનݚڀऀϓϩϑΝΠϧྨࣅࢉग़ςΩετಛ
5'*%'ʹجͮ͘ख๏ <4VHJZBNB +$%-`>• จͷಛநग़5'*%'ɺݚڀऀϓϩϑΝΠϦϯά5'Λ༻͍Δɻ• ޙऀจॻͱ୯ޠ͕গͳ͍͜ͱͰ*%'ʹѱ͞Λ͢ΔͨΊɻ• ۀ͕ຊͷΈͷएखͱͦΕҎ֎ͷγχΞͰϓϩϑΝΠϦϯάํ๏ΛมߋɻएखɿࣗͷۀͱͦͷࢀߟจݙΛ༻͍ΔɻγχΞɿࣗͷۀͱͦͷࢀߟจݙʹՃ͑ͯࣗͷۀΛҾ༻͍ͯ͠ΔจݙΛ༻͍Δɻ• จͱݚڀऀϓϩϑΝΠϧؒͷίαΠϯྨࣅΛࢉग़͠ɺ্ҐͷจΛਪનɻ, 4VHJZBNBBOE.: ,BO l4DIPMBSMZQBQFSSFDPNNFOEBUJPOWJBVTFSbTSFDFOUSFTFBSDIJOUFSFTUT z JO1SPD+$%- QQr …Ҿ༻ Ҿ༻࠷ۙͷจ΄ͲॏΈΛେ͖͘Ҿ༻एख γχΞ
จਪનͷݚڀ͕ٸܹʹ૿Ճ <"MJ 4DJFOUPNFUSJDT >• ಛʹҎ߱จຒΊࠐΈάϥϑຒΊࠐΈʢޙड़ʣؔ࿈ͷख๏͕ΜʹఏҊ͞Ε͍ͯΔɻ• ͜ͷจʮDJUBUJPOSFDPNNFOEBUJPOʯΛΫΤϦͱͯ͠จΛऩू͕ͨ͠ɺʮQBQFSSFDPNNFOEBUJPOʯؚΊΔͱڪΒͬ͘ͱଟ͍ɻ;"MJ *6MMBI ",IBO "6+BO BOE,.VIBNNBE l"OPWFSWJFXBOEFWBMVBUJPOPGDJUBUJPOSFDPNNFOEBUJPONPEFMT z4DJFOUPNFUSJDT WPM QQr ࠨਤͷ͏ͪਂֶशܥͷจlDJUBUJPOSFDPNNFOEBUJPOzͷจ
4DJ#&35 <#FMUBHZ&./-1`>• $4ɾੜҩֶʹ͓͚ΔສຊͷจϑϧςΩετͰ#&35Λֶशɻ• ΦϦδφϧ#&35ͷޠኮͱ4DJ#&35ͷޠኮͷΦʔόʔϥοϓɻ• ՊֶͱҰൠͷςΩετͷؒͰ୯ޠͷग़ݱසʹେ͖ͳҧ͍͕͋Δ͜ͱΛࣔͨ͠ɽ4DJ#&35#&35*#FMUBHZ FUBM l4DJ#&351SFUSBJOFE-BOHVBHF.PEFMGPS4DJFOUJGJD5FYU z1SPD&./-1 • ͳ͓#&35จॻͷจ຺ͷΈΛֶशʹ༻͍ΔͨΊɺҟͳΔจॻؒͷؔ࿈ੑΛଊ͑ΔΑ͏ʹ܇࿅͞Ε͍ͯͳ͍ɻ
จΤϯίʔμͷൺֱ <.PIBNFE)BTTBO 3FD4ZT`>• จਪનʹ͓͍ͯ#&35 &-.P 64& *OGFS4FOU 4DJ#&35Λൺֱɻ• #.ͰϥϯΩϯάީิΛߜ͔ͬͯΒΤϯίʔμͰϦϥϯΩϯά͢Δ߹ɼυϝΠϯಛԽܕͷ4DJ#&35Ͱͳ͘8JLJQFEJBؚΉҰൠจॻͰֶशͨ͠64&ͷํ͕ੑೳ͕ྑ͔ͬͨɻ.PIBNFE)BTTBO FUBM l#&35 &-.0 64&BOE*OGFS4FOUTFOUFODFFODPEFST5IFQBOBDFBGPSSFTFBSDIQBQFSSFDPNNFOEBUJPO z *O 1SPDPG"$.3FD4ZT-BUF#SFBLJOH3FTVMUTDPMPDBUFEXJUIUIFUI"$.$POGFSFODFPO3FDPNNFOEFS4ZTUFNT WPM QQr
41&$5&3<$PIBO "$-`>• 4DJ#&35ΛจਪનͳͲԠ༻͢ΔͨΊʹʮҾ༻ؔͷ༧ଌʯʹؔ͢Δ5SJQMFUϩεΛಋೖɻ• Ҿ༻͕ؔ༧ଌͰ͖ΔͳΒจؒͷؔ࿈ੑΛֶशͰ͖͍ͯΔͩΖ͏ͱ͍͏ߟ͑ɻ• ສͷֶश༻USJQMFUTͱສͷݕূ༻USJQMFUTΛ༻͍ͯ5JUBO7(16Ͱֶशɻ• ఔͰֶशྃɻ• طଘͷݴޠϞσϧʹൺͯෳλεΫͰߴ͍ੑೳΛࣔͨ͠ɻҾ༻ຒΊࠐΈઌͰۙ͘ͳΔΑ͏ʹ ຒΊࠐΈઌͰԕ͘ͳΔΑ͏ʹ"$PIBOFUBM l41&$5&3%PDVNFOUMFWFM3FQSFTFOUBUJPO-FBSOJOHVTJOH$JUBUJPOJOGPSNFE5SBOTGPSNFST z1SPD"$-
จਪનख๏ͷΧςΰϥΠζڠௐϑΟϧλϦϯάख๏ ίϯςϯπϕʔεख๏άϥϑຒΊࠐΈख๏ ϋΠϒϦουϢʔβͷڵຯ จͷ༰4JNJMBS4JNJMBSҾ༻ Ҿ༻
ֶज़ใ͔Βߏங͞ΕΔάϥϑ• ҟͳΔछྨͷϊʔυΤοδ͕ࠞࡏ͢ΔωοτϫʔΫͷ͜ͱΛҟछใωοτϫʔΫIFUFSPHFOFPVTJOGPSNBUJPOOFUXPSL)*/ͱΑͿɻ• ֶज़ใͷ߹ஶऀɺจɺॴଐɺձٞɺॴͳͲͷϊʔυ͕ߟ͑ΒΕΔɻֶज़ใͷ)*/ΤοδॏΈ͕શͯͰΤοδʹϥϕϧ͕͘ͳΒφϨοδάϥϑʹ૬ژλϫʔҰൠతͳφϨοδάϥϑͷྫॴJTBژࢢJTMPDBUFEJO"MJDFWJTJUFEJTBࢢ
3%'7FD<3JTUPTLJ *48$`>• ηϚϯςΟοΫΣϒͰΑ͘༻͍ΒΕΔɻ• φϨοδάϥϑ্ͰˠΤοδˠˠΤοδʜͷΑ͏ʹάϥϑΥʔΫ͠ɺύεΛ୯ޠͷܥྻมɻXPSEWFDΛద༻ͯ͠ϊʔυͷຒΊࠐΈΛಘΔɻ1FUBS 3JTUPTLJ BOE)FJLP1BVMIFJN l3%'7FD3%'(SBQI&NCFEEJOHTGPS%BUB.JOJOH z*OUFSOBUJPOBM4FNBOUJD8FC$POGFSFODF ݚڀऀ"ࣥචจൃදઌࠃࡍձٞB…
7013FD <,POH *&&&5&5$ >• ςΩετಛͱωοτϫʔΫߏʹج͖ͮจϊʔυͷຒΊࠐΈΛֶशɻ• ֤จͷςΩετͱҾ༻άϥϑΛͦΕͧΕXPSEWFD TUSVDWFDͰϕΫτϧԽɻ• ͦΕͧΕͷϕΫτϧۭؒͰۙͱͳΔจͱͷΤοδΛݩͷҾ༻άϥϑʹՃ͢Δɻ• άϥϑΥʔΫͱ4LJQHSBNͰϊʔυຒΊࠐΈΛಘΔɻ9,POH ..BP 88BOH +-JVBOE#9V 7013FD7FDUPS3FQSFTFOUBUJPO-FBSOJOHPG1BQFSTXJUI5FYU*OGPSNBUJPOBOE4USVDUVSBM*EFOUJUZGPS3FDPNNFOEBUJPO *&&&5SBOT&NFSHJOH5PQJDTJO$PNQVUJOH WPM OP QQ
0"(#&35<-JV ,%%`>• 0QFO"DBEFNJD(SBQIͷҟछใωοτϫʔΫΛςΩετԽͯ͠#&35Λֶशɻ• શจΛ༻͍ͨࣄલֶशͷ࣌ͰλΠτϧͱΞϒετϥΫτͷจͷؒʹஶऀ໊ͷτʔΫϯΛೖΕ͓ͯ͘ɻ• τʔΫϯʹϊʔυͷछྨͷΠϯσοΫεΛ༩ɻྫɿςΩετɺஶऀɻ• %PDWFD4DJ#&35ʹൺͯจਪનੑೳͷ্Λୡɻ9-JV %:JO +;IFOH 9;IBOH 1;IBOH ):BOH :%POH BOE+5BOH l0"(#&355PXBSETB6OJGJFE#BDLCPOF-BOHVBHF.PEFMGPS"DBEFNJD,OPXMFEHF4FSWJDFT zJO1SPD,%%` QQr
จਪનͷධՁํ๏• େ͖͚ͯ࣍͘ͷͭͷΞϓϩʔν͕͋Δɻ࣮ࡍʹਓʹධՁͯ͠Β͏ɻϢʔβධՁΣϒϓϥοτϑΥʔϜʹͯϢʔβͷڍಈΛධՁɻΦϯϥΠϯධՁHSPVOEUSVUIΛઃఆͯ͠ਖ਼ղͳͲΛࢉग़ɻΦϑϥΠϯධՁ
ϢʔβධՁ• ਪન݁ՌΛϢʔβʹఏࣔ͠ɺ߲͝ͱʹΛ͚ͭͯΒ͏ɻ• ৽نੑɺ༗༻ੑɺख़࿅ऀ͚͔Ͳ͏͔ɺFUDྫʣ<3BBNLVNBS *1. >"43BBNLVNBS 4'PP BOE/1BOH l6TJOHBVUIPSTQFDJGJFELFZXPSETJOCVJMEJOHBOJOJUJBMSFBEJOHMJTUPGSFTFBSDIQBQFSTJOTDJFOUJGJDQBQFSSFUSJFWBMBOESFDPNNFOEFSTZTUFNT z*OGPSNBUJPO1SPDFTTJOHBOE.BOBHFNFOU WPM OP QQr • ਓతίετ͕େ͖͍ɻಛʹ༷ʑͳͷධՁऀΛूΊΔͷࠔɻ• ධՁऀͷόοΫάϥϯυεΩϧ͕Өڹ͢Δ໘ɻü ֶੜ໊ͱڭ৬һ໊ʹϝʔϧͰ࣮ݧڠྗΛґཔɻࢀՃ݅ʮҰຊҎ্ͷจΛࣥචͨ͜͠ͱʯɻü ܭʹର͠ஈ֊ͷϦοΧʔτईͰճɻü ֶੜͷํ͕ڭ৬һΑΓਪનʹؔ͢Δຬ͕༏Ґʹߴ͍͜ͱΛࣔͨ͠ɻ
ΦϯϥΠϯධՁ• ϨʔςΟϯάΛूΊΔ͜ͱͳ͘ਪનγεςϜར༻ऀͷڍಈΛఆྔԽ͢Δɻ• &ίϚʔεͰͷਪનΞϧΰϦζϜධՁͰΑ͘༻͍ΒΕΔํ๏ɻ• ΫϦοΫεϧʔʜਪન͞Εͨͷͷ͏ͪɺ࣮ࡍʹΫϦοΫ͞Εׂͨ߹ɻϢʔβͷຬͱ૬͕ؔ͋Δɻ• μϯϩʔυɺߪೖ ͳͲ• ϓϥοτϑΥʔϜͷҧ͍͕༩͑ΔӨڹΛݕূ͢Δඞཁ͕͋Δɻ• ࣮γεςϜ͕ඞཁͳͨΊֶज़ใਪનͷͰಛఆάϧʔϓ͔͠ߦ͍ͬͯͳ͍ɻྫʣ<$PMMJOT+$%-`>จݙਪનɾཧͷΦʔϓϯιʔεϓϩδΣΫτʹͯສຊͷจݙ͔ΒϢʔβʹਪનɻ࣮ࢪɻTF-IDF(95%)Doc2vec (2%)キーフレーズ(3%)ਪનͷΞϧΰϦζϜ༁ϓϥοτϑΥʔϜʹΑͬͯΞϧΰϦζϜͷੑೳॱҐ͕มಈͨ͠ʢʂʣ
ΦϑϥΠϯධՁ• จσʔληοτͷҾ༻ؔΛٖࣅతʹHSPVOEUSVUIͱͯ͠औΓѻ͏ɻ• ਪનͨ͠จ͕ΫΤϦจʹҾ༻͞Ε͍ͯΕਖ਼ྫɺ͞Ε͍ͯͳ͚Εෛྫɻ• ϢʔβධՁΑΓਓతίετ͕͍ͨΊ΄ͱΜͲͷݚڀ͕͜ͷධՁํ๏Λ࠾༻ɻྫʣ%#-1σʔληοτ ຊͷจ ݸͷҾ༻ؔྫʣ"$-"OUIPMPHZ/FUXPSL• ʮຊͳΒҾ༻͓͔ͯ͘͠͠ͳ͍จʯ͕ෆਖ਼ղʹͳͬͯ͠·͏ɻIUUQTXXXBNJOFSPSHDJUBUJPO ຊͷจ ݸͷҾ༻ؔIUUQTBBOIPXEPXOMPBEʜͲͷධՁํ๏ҰҰɺ͞ΒͳΔݕ౼͕ඞཁɻ
࠷ۙͷݚڀɿΞεϖΫτผͷจຒΊࠐΈ• 4PMVUJPO5BJMPS <5BLBIBTIJ,BUTVSBJ &$*3`>• λεΫ͕ࣅ͍ͯΔ͕ख๏͕ҟͳΔจΛਪનɻ• 4QFDJBMJ[FE%PDVNFOU&NCFEEJOHT <0TUFOEPSG +$%-`>• จͷλεΫɺख๏ɺσʔληοτʹؔ͢Δ෦͝ͱʹຒΊࠐΈΛࢉग़͢Δɻ• 1BQFST8JUI$PEFͷ UBTL NFUIPE EBUBTFUʹؔ͢Δهड़ΛͱʹUSJQMFUTΛ࡞ɻจ จಉ͡σʔληοτΛ͍ͬͯΔˠۙ͘ʹஔ"ͷจ #ͷจᶃΞϒετϥΫτͷྨࣅᶄจɾղܾจͷྨࣅྨࣅͷߴ͍ॱʹਪનจղܾจ#&35Λ༻͍ͯΞϒετͷจΛഎܠɾతɾख๏ʹྨ. 0TUFOEPSGG FUBM l4QFDJBMJ[FEEPDVNFOUFNCFEEJOHTGPSBTQFDUCBTFETJNJMBSJUZPGSFTFBSDIQBQFST z JO1SPD+$%-"SUJDMF QQr 55BLBIBTIJBOE.,BUTVSBJ l4PMVUJPO5BJMPS4DJFOUJGJD1BQFS3FDPNNFOEBUJPO#BTFEPO'JOF(SBJOFE"CTUSBDU"OBMZTJT zJO&$*3 QQr
·ͱΊͱࠓޙͷల• ֶज़ใݕࡧͱਪનʹؔͯ͠ɺಛʹจΛରͱͨ͠ݚڀΛհͨ͠ɻ• ਪનΦϑϥΠϯධՁ͕ओྲྀ͕ͩɺਖ਼ղσʔλͷ࡞Γํʹ͞ΒͳΔݕ౼͕ඞཁɻ• શจ͕͑ΔจσʔλϕʔεݶΒΕ͓ͯΓɺ͔ͭԣஅతͳֶज़ใਪનʹؔ͢Δݚڀൃల్্ͱ͍͑Δɻ• ࣗવݴޠॲཧʹΑΔΞεϖΫτੳࠓޙ͞Βʹల։͍ͯ͘͠ͱ༧ɻڞಉݚڀͳͲ͓ؾܰʹ͓͕͚͍ͩ͘͞ʂؔ࿈ݚڀऀਪન ݚڀτϨϯυੳ ஶऀಉఆใܥཧܥҩֶܥը૾ҩྍϩϘοτͦͷଞͷݚڀτϐοΫ