Spacyでお手軽NLP / NLP with spacy

Slide 1

Slide 1 text

Ͱ͓खܰ/-1 )JSBNBUTV!ϨτϦόηϛφʔ ը૾IUUQTHJUIVCDPNFYQMPTJPOTQB$ZCMPCNBTUFSXFCTJUFBTTFUTJNHMPHPTWH

Slide 2

Slide 2 text

Tsukuba, M2, NLP himkt

Slide 3

Slide 3 text

5-%3 w 1ZUIPOͷࣗવݴޠॲཧϥΠϒϥϦͰ͋Δ4QB$Zͷ঺հ w Wͷ͓΋͠Ζػೳʹ͍ͭͯ w 4QB$ZͰ೔ຊޠςΩετΛॲཧ͢Δ

Slide 4

Slide 4 text

"CPVU4QB$Z ը૾IUUQTTQBDZJP https://spacy.io

Slide 5

Slide 5 text

IUUQTTQBDZJP "CPVU4QB$Z “Industrial-Strength NLP” w /POEFTUSVDUJWFUPLFOJ[BUJPO w /BNFEFOUJUZSFDPHOJUJPO w 4VQQPSUGPSMBOHVBHFT w TUBUJTUJDBMNPEFMTGPSMBOHVBHFT w ʜFUD IUUQTHJUIVCDPNFYQMPTJPOTQB$Z

Slide 6

Slide 6 text

#BTJDVTBHFPG4QB$Z

Slide 7

Slide 7 text

.BOZ'FBUVSFT ը૾IUUQTTQBDZJPVTBHFGBDUTpHVSFT

Slide 8

Slide 8 text

'BTUFTUJOUIFXPSME w %FQFOEFODZQBSTFSͷύϑΥʔϚϯεൺֱ w ࿦จIUUQTBDMXFCPSHBOUIPMPHZ111QEG w ࿦จʹ͋Δͷ͸41&&%ͷදͰɼ"DDVSBDZ͸ࣗલͰ࡞ΒΕͨ΋ͷʁ w $IPJͷϕϯνϚʔΫ࣌ʹ͸TQB$ZW͸ະϦϦʔεͳͷͰOB ը૾IUUQTTQBDZJPVTBHFGBDUTpHVSFT

Slide 9

Slide 9 text

4QFFEDPNQBSJTPOXJUIPUIFSMJCSBSJFT w ଞͷࣗવݴޠॲཧϥΠϒϥϦͱͷ଎౓ൺֱ w ࿦จͰ͸ͳ͘࡞ऀ͕ௐࠪͨ͠΋ͷ DPOEVDUFEJO w ϦϙδτϦIUUQTHJUIVCDPNFYQMPTJPOTQBDZCFODINBSLT ը૾IUUQTTQBDZJPVTBHFGBDUTpHVSFT

Slide 10

Slide 10 text

.PEFMDPNQBSJTPO w ݴޠʹΑͬͯ͸ෳ਺ͷαΠζͷϞσϧ͕͋Δ FO GS FT w 104UBHHFS/&3UBHHFS%FQFOEFODZQBSTFS w 8JUIPVUBOZQSFQSPDFTTJOH EBUBTFU ը૾IUUQTTQBDZJPVTBHFGBDUTpHVSFT

Slide 11

Slide 11 text

/-5,BOE4QB$Z w /-5,P⒎FSTTPNFPGUIFTBNFGVODUJPOBMJUZBTTQB$Z w *ODPNQBSJTPOUPTQB$Z /-5,UBLFTBNVDINPSF  CSPBEDIVSDIBQQSPBDI w TQB$ZJTBMTPNVDINPSFQFSGPSNBODFGPDVTTFEUIBO/-5, XIFSFUIFUXPMJCSBSJFTQSPWJEFUIFTBNFGVODUJPOBMJUZ TQB$ZTJNQMFNFOUBUJPOXJMMVTVBMMZCFGBTUFSBOENPSF BDDVSBUF Ҿ༻IUUQTTQBDZJPVTBHFGBDUTpHVSFT

Slide 12

Slide 12 text

0UIFSMJCSBSJFTBOETQB$Z 1ZUPSDIIUUQTHJUIVCDPNQZUPSDIQZUPSDICMPCNBTUFSEPDTTPVSDF@TUBUJDJNHQZUPSDIMPHPEBSLTWH "MMFO/-1IUUQTHJUIVCDPNBMMFOBJBMMFOOMQCMPCNBTUFSEPDTUBUJDBMMFOOMQMPHPEBSLQOH (FOTJNIUUQTHJUIVCDPN3B3F5FDIOPMPHJFTHFOTJNCMPCEFWFMPQEPDTTSDSFBENF@JNBHFTSBSFQOH $V1ZIUUQTHJUIVCDPNDVQZDVQZCMPCNBTUFSEPDTJNBHFDVQZ@MPHP@QYQOH JOUPSDIUFYU GPS(16BDDFMFSBUJPO XPSEWFDUPS QJQFMJOF

Slide 13

Slide 13 text

4QB$ZWͷݸਓతʹ޷͖ͳػೳ w EJTQMB$ZͰ͌͢΀Εʔ͠ʔʁ w 4QB$ZͷՄࢹԽϞδϡʔϧ w ͖Ε͍Ͱ͍͍ײ͡ͳը૾Λ࡞ͬͯ͘ΕΔ w ݻ༗දݱநग़ͱ܎Γड͚ղੳͷ݁ՌΛՄࢹԽͯ͘͠ΕΔ w 4QB$ZͰղੳͨ͠ΦϒδΣΫτΛͦͷ··࢖͑Δ w 47(ܗࣜͷը૾͕ग़ྗ͞ΕΔ

Slide 14

Slide 14 text

8FCαΠτ্ͷ/&3ͷσϞ ը૾IUUQTFYQMPTJPOBJEFNPTEJTQMBDZFOU

Slide 15

Slide 15 text

4QB$ZY+VQZUFS/PUFCPPL

Slide 16

Slide 16 text

4QB$ZBOE೔ຊޠ w WͰͷ೔ຊޠରԠ 13 w ೔ຊޠܗଶૉղੳث+BOPNFΛϥοϓ͢ΔܗͰ࣮૷ w WͰͷܗଶૉղੳثҠߦ *TTVF 13 w ೔ຊޠ6OJWFSTBM%FQFOEFODZσʔλ͸6OJ%JDͰ෼ׂ͞Ε͍ͯΔ w +BOPNF͸ݱࡏͷͱ͜Ζ6OJ%JDʹະରԠ w .F$BCʹҠߦ

Slide 17

Slide 17 text

4QB$ZBOE೔ຊޠ

Slide 18

Slide 18 text

4QB$ZBOE೔ຊޠ "OTXFSVTF6OJ%JD

Slide 19

Slide 19 text

4QB$ZBOE೔ຊޠ

Slide 20

Slide 20 text

೔ຊޠ/&3%FQFOEFODZQBSTJOHXJUI4QB$Z w ݁࿦ݱࡏ͸Ͱ͖ͳ͍ w 4QB$Zʹ͸ࣗ෼Ͱ5BHHFS΍1BSTFSΛֶशͰ͖Δ  ػߏ͕૊Έࠐ·Ε͍ͯΔ w ೔ຊޠ6OJWFSTBM%FQFOEFODJFT͸ެ։͞Ε͍ͯΔ  IUUQTHJUIVCDPN6OJWFSTBM%FQFOEFODJFT6%@+BQBOFTF(4%ͳͲ Ϟσϧࣗ࡞Ͱ͖ΔͷͰ͸ʂʁ

Slide 21

Slide 21 text

"EEJOH-BOHVBHFTVQQPSU 4QB$ZͷݴޠϞδϡʔϧͷ  ίϯϙʔωϯτ ը૾IUUQTTQBDZJPVTBHFBEEJOHMBOHVBHFTTFDUJPOUSBJOJOH

Slide 22

Slide 22 text

"EEJOH-BOHVBHFTVQQPSU ը૾IUUQTTQBDZJPVTBHFBEEJOHMBOHVBHFTTFDUJPOUSBJOJOH 1PXFSFECZ.F$BC  4QB$ZͷֶशϞδϡʔϧ͕࢖͑ͳ͍ʁ

Slide 23

Slide 23 text

೔ຊޠTQB$Zͷݱࡏͷ໰୊ʁ w ࣙॻ͕6OJ%JDͰ͋Δ͜ͱ͕૝ఆ͞Ε͍ͯΔ JTTVF w ʢ͓ͦΒ͘ʣଟ͘ͷ؀ڥͰ͸.F$BCͷσϑΥϧτࣙॻ͸*1"EJD w <50%0>+BQBOFTF5PLFOJ[FSͰ5BHHFSΛ࡞͍ͬͯΔ͕ɼ  ͜ͷίϯετϥΫλͷҾ਺͸֎෦͔Β৮Εͳͦ͏ w ൃԻ͕ະొ࿥ͷ୯ޠΛղੳ͢ΔͱΤϥʔ *1"EJD 13 w ࣙॻ͝ͱͷग़ྗͷࠩҟΛٵऩ͢ΔΠϯλʔϑΣʔε͕ඞཁʁ w /&3΍%FQFOEFODZ1BSTJOHͷϞσϧ͸·ͩଘࡏ͠ͳ͍ w ݱࡏ͸A"MQIBUPLFOJ[BUJPOTVQQPSUA w ෼͔ͪॻ͖͕ඞཁͳݴޠͷରԠ͸Ͳ͏͢Ε͹ʜ w தࠃޠ΋ࣅͨঢ়گͰࢭ·͍ͬͯΔ 13

Slide 24

Slide 24 text

͜ΕΛ΍Ε͹ͱΓ͋͑ͣςετ͸Ͱ͖ͦ͏ USBWJTͰ.F$BCΛΠϯετʔϧ͢ΔΑ͏ʹ͢Δ  TQBDZNPEFMTʹ6OJ%JDΛొ࿥͢Δ  QZUIPONTQBDZEPXOMPBEKBΛ੔උ͢Δ /&3ͱ%FQFOEFODZ1BSTJOH͸͋ͱͰௐ΂Δʜ

Slide 25

Slide 25 text

·ͱΊ w ࣗવݴޠॲཧϥΠϒϥϦTQB$Zͷ঺հ w *OEVTUSZൃͷࣗવݴޠॲཧϥΠϒϥϦ w WͰೖͬͨՄࢹԽϞδϡʔϧ͕͍͍ײ͡ w EJTQMB$ZΛ࢖͍͍ͬͨײ͡ͳՄࢹԽ w TQB$ZͰͷ೔ຊޠςΩετॲཧ͸·ͩෆ׬શ w ݱঢ়ܗଶૉղੳͷΠϯλʔϑΣʔε QJQJOTUBMMTQBDZ 3VO5IJTDPNNBOE