Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Speaker Deck
PRO
Sign in
Sign up
for free
COVID-19 に対するオープンデータの取り組み
Asei Sugiyama
April 25, 2020
Technology
0
110
COVID-19 に対するオープンデータの取り組み
第44回 Machine Learning 15minutes! で登壇した際の資料です
Asei Sugiyama
April 25, 2020
Tweet
Share
More Decks by Asei Sugiyama
See All by Asei Sugiyama
asei
2
110
asei
0
490
asei
0
110
asei
0
570
asei
0
450
asei
0
1.1k
asei
1
540
asei
0
400
asei
0
85
Other Decks in Technology
See All in Technology
yunoda
0
140
uzabase_saas_product
0
100
tenjuu99
1
300
hiroyaiizuka
0
160
dena_tech
0
170
htomine
0
170
900groove
2
530
mahito
0
240
akitok_
2
780
aditya45
2
2.4k
kaniza
0
120
sadayoshitada0919
0
370
Featured
See All Featured
chriscoyier
498
130k
lara
15
2.7k
philhawksworth
190
17k
jensimmons
207
10k
62gerente
587
200k
tanoku
86
8.6k
tenderlove
53
3.5k
searls
204
36k
holman
447
130k
hursman
106
9.3k
chriscoyier
779
240k
chrislema
173
14k
Transcript
COVID-19 ʹؔ͢Δ ΦʔϓϯσʔλͷऔΓΈ ୈ 44 ճ Machine Learning 15minutes! 1
ࣗݾհ • ਿࢁ Ѩ • Software Engineer @ Repro AI
Labs • TensorFlow ίϯτϦϏϡʔλʔ • TFX : Issue ཱͯͨΓ PR ཱͯͨΓ • docs-l10n : ༁ & ϨϏϡʔ • ػցֶशਤؑ ڞஶ 2
ҙࣄ߲ • ͋͘·ͰݸਓతͳݟղͰ͋Γɺॴଐ͢Δ͍͔ͳΔஂମͱແؔ Ͱ͢ • σʔλαΠΤϯεͷ͋·Γ͠·ͤΜ • ػցֶशͷ͠·ͤΜ • COVID-19
ʹؔ͢Δকདྷ༧ଌͷ͠·ͤΜ • COVID-19 ͷײછ༧ͷ͠·ͤΜ 3
ຊͷ༰ 1.Open Data ͱ <- 2.COVID-19 ʹؔ͢Δ Open Data ͱͦͷ׆༻
3.զʑσʔλͰԿ͕Ͱ͖Δͷ͔ 4
Open Data ͱ • “Open data and content can be
freely used, modified, and shared by anyone for any purpose” from The Open Definition • ϚγϯϦʔμϒϧͰ͋Δ͜ͱॏཁࢹ • Data.gov : ΞϝϦΧͷΦʔϓϯσʔ λͷαΠτ (ӈਤ) • Data Portals : 588 ͷΦʔϓϯ σʔλͷΧλϩά 5
ຊͷ Open Data • ࣏ࣗମ͕ؤு͍ͬͯΔ • data.go.jp • σʔλγςΟḉߐϙʔλϧαΠτ •
͍͢͝ (ޙड़) • ౦ژΦʔϓϯσʔλΧλϩάαΠτ ϗʔϜϖʔδ 6
ΦʔϓϯσʔλΛ͡ΊΑ͏1 • ಠཱߦ๏ਓɺ࣏ࣗମͳͲ͕อ ༗͢Δެڞσʔλ͕ɺࠃຽاۀͳͲ ͷར׆༻͞Ε͍͢Α͏ʹػձಡʹ దͨ͠ܗͰɺೋ࣍ར༻Մೳͳϧʔϧͷ ͱͰެ։͞ΕΔ͜ͱɺ·ͨɺͦͷΑ ͏ʹެ։͞ΕͨσʔλΛࢦ͠·͢ 1 http://www.kantei.go.jp/jp/singi/it2/densi/kettei/
opendata_tebikisyo.pdf 7
σʔλγςΟḉߐϙʔ λϧαΠτ • ͞·͟·ͳσʔλΛ PDF Ͱͳ͘ CSV, XML, RDF, Linked
RDF Ͱެ ։͍ͯ͠Δ • ϥΠηϯε CC BY • ྉཧϨγϐ (XML) 8
9
͑͞࢈ࡊͷྉཧϨγϐ (ൈਮ) ݟ͢͞ͷͨΊɺ࠷ޙͷߦʹ͚ͩվߦΛՃ͑ͨ <http://linkdata.org/resource/rdf1s8262i#2> <http://www.w3.org/2000/01/rdf-schema#label> "2"@ja ; <http://linkdata.org/property/rdf1s8262i#%E6%96%99%E7%90%86%E5%90%8D> "΄͏ΕΜͱ͠ͷφϜϧ"@ja ;
<http://linkdata.org/property/rdf1s8262i#%E4%BA%BA%E6%95%B0> "4ਓ"@ja ; <http://linkdata.org/property/rdf1s8262i#%E6%9D%90%E6%96%991> "͠ 170gʢ2/3ାʣ"@ja ; <http://linkdata.org/property/rdf1s8262i#%E6%9D%90%E6%96%99%EF%BC%92> "΄͏ΕΜ 130gʢ2/3ଋʣ"@ja ; <http://linkdata.org/property/rdf1s8262i#%E6%9D%90%E6%96%99%EF%BC%93> "͕ܲΒεʔϓͷૉ খ͞͡2"@ja ; <http://linkdata.org/property/rdf1s8262i#%E6%9D%90%E6%96%99%EF%BC%94> "͝·༉ খ̍͞͡"@ja ; <http://linkdata.org/property/rdf1s8262i#%E8%A9%B3%E7%B4%B0URL> <https://www.city.sabae.fukui.jp/kenko_fukushi/hoken_kenkozukuri/mugenyasai.files/receipi.pdf> . 10
͑͞࢈ࡊͷྉཧϨγϐ • http://linkdata.org/work/ rdf1s8262i • ߹ܭμϯϩʔυ 4 ճ (2020-04-23 ࣌)
• ࠷ऴߋ৽ : 2020/03/17 • Text, Excel, RDF ܗࣜ 11
ຊͷ༰ 1.Open Data ͱ 2.COVID-19 ʹؔ͢Δ Open Data ͱͦͷ׆༻ <-
3.զʑσʔλͰԿ͕Ͱ͖Δͷ͔ 12
COVID19 ʹؔ͢Δ Open Data ͱͦͷ׆༻ ٸʹσʔλެ։ɾ׆༻ͷಈ͖͕͕͍ͬͯΔɻ • ౦ژ ৽ܕίϩφΠϧεײછରࡦαΠτ •
৽ܕίϩφΠϧεରࡦμογϡϘʔυ 13
౦ژ ৽ܕίϩφ ΠϧεײછରࡦαΠ τ • ౦ژΦʔϓϯσʔλΧλϩάαΠτͷ σʔλ͔Β࡞ • αΠτࣗମ MIT
ϥΠηϯε • Code for Japan ͷ׆ಈͷ 1 ͭ 14
15
StopCovid19 શࠃ൛ • ౦ژͷΦʔϓϯιʔεΛར༻ͨ͠αΠ τͷ·ͱΊαΠτ • ͜ͷαΠτࣗମ MIT ϥΠηϯε •
codeforjapan/stopcovid19 16
৽ܕίϩφΠϧεର ࡦμογϡϘʔυ • ହհ (@taisukef) ࢯ • COVID-19 ͷױऀ͕ར༻Մೳͳපচ ʹணͨ͠αΠτ
• ްੜ࿑ಇলఏڙͷ৽ܕίϩφΠϧε ཅੑױऀ PDF σʔλΛ JSON-API Խͨ݁͠ՌΛར༻ • ιʔεCC BYϥΠηϯεͰఏڙ͞Εͯ ͍Δ 17
ಓݝผ৽ܕίϩφΠϧεײછऀϚοϓ • δϟοάδϟύϯגࣜձࣾ • σʔλ CC-BY-NC 4.0 • ༷ʑͳσʔλιʔε͔Βूܭ͍ͯ͠Δ ༷
(ෳਓʹΑΔ࡞ۀ) • ՄࢹԽ ArcGIS (ESRI) 18
தͷਓͷ • https://twitter.com/ oohamazaki/status/ 1253306592595832835?s=20 19
SIGNATE COVID-19 νϟϨϯδ • ϑΣʔζ 1. σʔληοτͷߏங • ϑΣʔζ 2.
σʔλੳ • ϑΣʔζ 3. ጶױऀ༧ଌ 20
21
ຊͷ༰ 1.Open Data ͱ 2.COVID-19 ʹؔ͢Δ Open Data ͱͦͷ׆༻ 3.զʑσʔλͰԿ͕Ͱ͖Δͷ͔
<- 22
ཁࢫ • σʔλ͕࣍ʑʹެ։͞Ε͍ͯΔ͕ɺඇઐՈͰ͋ΔҰൠࢢຽ͕ ݟͯಘΒΕΔͷগͳ͍ • ૬ஊ݅ͷരൃతͳ૿େ͕ͳ͍ͱ͍͏͜ͱɺ૭ޱ͕ർฐͯ͠ ͍ΔՄೳੑΛࣔ͢ • Ұൠࢢຽͱͯ࣍͠ʹ Q&A
Λඋ͠ɺͦΕΛΞΫηε͍͢͠ ܗͰఏڙͰ͖ΔΑ͏ʹ͢Δ͜ͱΛఏҊ͍ͨ͠ 23
ݸਓతͳ҆৺ • COVID-19 ʹؔ͢Δ AI ສೳ͕ग़ͯ ͍ͳ͍͜ͱ • ༧ଌϞσϧʹΑΔ༬ݴऀ͕࣍ʑʹݱΕ Δঢ়گͰͳ͍ͷෆதͷ͍
• σʔληοτͷඋͱ͍͏ͬͱॏ ཁͳՕॴʹϦιʔεׂ͕͔Ε͍ͯΔ 24
ػցֶशʹཱͭͷ͔ʁ • ػցֶश·Ͱͷεςοϓʹ͓͍ͯɺݱࡏۀͷఆٛͷஈ֊ 1.ۀ͕ະఆٛ 2.ۀ͕ఆٛ͞ΕɺखॱॻΛ࡞Δ 3.खॱ͕໌֬ʹͳΓɺγεςϜԽΛߦ͏ <- ࠓ͜ͷ͋ͨΓ 4.ใΛՄࢹԽ͠ɺҙࢥܾఆΛߦ͏ <-
ઐՈ͜͜ 5.ҙࢥܾఆͷࣗಈԽ or ิॿ <- ػցֶश͜͜ 25
σʔλαΠΤϯςΟετʹཱͭͷ͔ʁ • σʔλͷ׆༻ʹඞཁͳεΩϧͷ • σʔλαΠΤϯε • ΤϯδχΞϦϯά • υϝΠϯࣝ •
ҰൠͷσʔλαΠΤϯςΟετɾΤϯδχΞҩֶެऺӴੜ ͷ͕ࣝඞͣ͠๛ͱ͍͏Θ͚Ͱͳ͍ 26
ࣗʹԿ͕Ͱ͖ΔͷͩΖ͏͔ʁ • COVID-19 ͷײછͷ֦େঢ়گʹ͍ͭͯूΊΒΕͨσʔλɺඇ ઐՈͷखʹ༨Δ • Titanic ͷσʔλ͔ΒɺզʑԿ͔༗༻ͳݟΛಘΒΕͨͩ Ζ͏͔ʁ •
υϝΠϯࣝΛൈ͖ʹɺཧϞσϧػցֶशϞσϧΛ༻͍ͯ কདྷ༧ଌΛͨ͠ͱ͜ΖͰɺͦͷ݁ՌԿ͔ҙຯ͕͋ΔͩΖ͏ ͔ʁ 27
Amazon (AWS) ͷύ ϒϦοΫσʔλϨΠΫ • σʔλʹΞΫηε͘͢͠ͳͬͨ • ϚΫϩͳҙࢥܾఆ͔֬ʹͰ͖Δ͔ ͠Εͳ͍ •
զʑ͜ΕΛݟͯԿΛ͢Ε͍͍Μͩ Ζ͏͔ʁ 28
See how your community is moving around differently due to
COVID-192 • Google ͷϨϙʔτ • Transit stations Workplace ΛݮΒͤྑͦ͞͏ͳײ͕͢͡Δ • ײછऀΑΓམͪண͍ͯݟΒΕΔ 2 Google LLC "Google COVID-19 Community Mobility Reports." https://www.google.com/covid19/mobility/ Accessed: 2020-04-24. 29
զʑ͕ݟΔ͖ࢦඪ ͳʹ͔ʁ • ࣍ʹΔ͜ͱ͕ݟ͑ΔࢦඪΛݟ͍ͨ • ͜ΕͲͷۀʹண͖͔͢ͱ͍͏ ʹಉ͡ • ͍߹Θͤۀʹண͢Δ͜ͱΛఏҊ ͍ͨ͠
• ౦ژ ৽ܕίϩφίʔϧηϯλʔ૬ஊ ݅ͷྦྷܭ͕ઢܗʹ͔͠৳ͼ͍ͯͳ͍ 30
31
࣍ʹඋ͖͢σʔλͳʹ͔ʁ • ඞཁͳͷ͍߹Θͤͱͦͷճ • ༷ʑͳରࡦ͕ଧͨΕ͍ͯΔҰํɺ૭ޱͷΞΫηε໌֬Ͱͳ ͍ (ཧతʹॴʹߦ͚ͳ͍ʂ) ͷͰ͜͏͍ͬͨ૭ޱʹؔͳ ͍͍߹Θ͕ͤࡴ౸͍ͯ͠Δ͜ͱ͕༧͞ΕΔ •
Q&A σʔλΛ༻͍ͨݕࡧΤϯδϯ໌Β͔ʹ༗ޮ • Q&A σʔλͱγφϦΦΛݩʹͨ͠νϟοτϘοτ༗ޮ 32
33
https://corona.go.jp/ ʹઃஔࡁ 34
ఆ͞ΕΔ՝ • ηϯγςΟϒͳ࣭͕૬ྔدͤΒΕ Δ͜ͱ͕ఆ͞Εɺੜσʔλͷอ Ͱ͖ͳ͍ • ঢ়گ͕·͙Δ͘͠มԽ͢Δ݁Ռɺಉ ࣭͡ʹରͯ͠దͳճ͕ৗʹม ΘΓଓ͚ΔͨΊɺσʔληοτͷߋ৽ ͕ඞཁ
• લऀͱ͔͘ޙऀʹ͍ͭͯࢢຽͰ Ͱ͖Δ͜ͱ͕͋ΔͷͰ 35
ͳͥզʑϚεΫΛٻ Ίͯ͠·͏ͷ͔ • ڴҖʹରͯ͠Ͱ͖Δ͜ͱ͕ͳʹͳ͍ ͱ͍͏ετϨε͔ΒճආͰ͖Δ • ϚεΫങ͍ٻΊΔ͜ͱ͕Ͱ͖Δͷ Ͱɺڻҟʹରͯ͠औΕΔΞΫγϣϯʹ ͳ͍ͬͯΔ http://www.osaka-kyoiku.ac.jp/
~toyosawa/klecture09.pdf 36
σʔληοτͷඋࢢຽ͕ߦ͑ΔΞΫγϣϯ • ඇઐՈͰࢀՃͰ͖Δ • ࡞Δ͜ͱΑΓอक͢Δ͜ͱʹଟେͳख͕͔͔ؒΔ (ϦϯΫ Εσʔλͷܗࣜมߋͷରॲͱ͔) • σʔληοτ͕͋ΕΤϯδχΞͷྗͰιϑτΣΞʹͰ͖Δ •
ιϑτΣΞʹͰ͖Ε֦ࢄͰ͖Δ • SIGNATE ͷऔΓΈ͋Δҙຯཧత 37
Recap • σʔλ͕࣍ʑʹެ։͞Ε͍ͯΔ͕ɺඇઐՈͰ͋ΔҰൠࢢຽ͕ ݟͯಘΒΕΔͷগͳ͍ • ૬ஊ݅ͷരൃతͳ૿େ͕ͳ͍ͱ͍͏͜ͱɺ૭ޱ͕ർฐͯ͠ ͍ΔՄೳੑΛࣔ͢ • Ұൠࢢຽͱͯ࣍͠ʹ Q&A
Λඋ͠ɺͦΕΛΞΫηε͍͢͠ ܗͰఏڙͰ͖ΔΑ͏ʹ͢Δ͜ͱΛఏҊ͍ͨ͠ 38