Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
AmazonPersonalize活用事例(cyma導入事例)_Ateam Inc.
Search
Ateam Inc.
July 17, 2019
Technology
0
1.5k
AmazonPersonalize活用事例(cyma導入事例)_Ateam Inc.
Ateam Inc.
July 17, 2019
Tweet
Share
More Decks by Ateam Inc.
See All by Ateam Inc.
エイチーム M&A方針説明会資料(2024年8月7日)
ateam
0
2.8k
Recruitment materials
ateam
2
25k
"Ateam People"
ateam
0
190k
"Ateam People"
ateam
0
1.8k
Other Decks in Technology
See All in Technology
[SRE kaigi 2025] ガバメントクラウドに向けた開発と変化するSRE組織のあり方 / Development for Government Cloud and the Evolving Role of SRE Teams
kazeburo
3
1.4k
ソフトウェアアーキテクトのための意思決定術: Software Architecture and Decision-Making
snoozer05
PRO
4
650
Redmineの意外と知らない便利機能 (Redmine 6.0対応版)
vividtone
0
140
インフラコストとセキュリティ課題解決のためのリアーキテクチャリング / srekaigi2025
hgsgtk
3
3.5k
Asset Centric な データ変換パイプラインの攻略法
recruitengineers
PRO
1
100
信頼性を支えるテレメトリーパイプラインの構築 / Building Telemetry Pipeline with OpenTelemetry
ymotongpoo
9
4.1k
2週に1度のビッグバンリリースをデイリーリリース化するまでの苦悩 ~急成長するスタートアップのリアルな裏側~
kworkdev
PRO
8
5.5k
RevOpsへ至る道 データ活用による事業革新への挑戦 / path-to-revops
pei0804
1
350
インシデントキーメトリクスによるインシデント対応の改善 / Improving Incident Response using Incident Key Metrics
nari_ex
0
2.8k
0→1事業こそPMは営業すべし / pmconf #落選お披露目 / PM should do sales in zero to one
roki_n_
PRO
2
4k
Microsoft Ignite 2024 最新情報!Microsoft 365 Agents SDK 概要 / Microsoft Ignite 2024 latest news Microsoft 365 Agents SDK overview
karamem0
0
160
Skip Skip Run Run Run ♫
temoki
0
310
Featured
See All Featured
The Pragmatic Product Professional
lauravandoore
32
6.4k
jQuery: Nuts, Bolts and Bling
dougneiner
63
7.6k
RailsConf & Balkan Ruby 2019: The Past, Present, and Future of Rails at GitHub
eileencodes
132
33k
How to Ace a Technical Interview
jacobian
276
23k
GitHub's CSS Performance
jonrohan
1030
460k
Faster Mobile Websites
deanohume
305
30k
Embracing the Ebb and Flow
colly
84
4.5k
KATA
mclloyd
29
14k
How GitHub (no longer) Works
holman
312
140k
VelocityConf: Rendering Performance Case Studies
addyosmani
327
24k
Java REST API Framework Comparison - PWX 2021
mraible
28
8.3k
Why Our Code Smells
bkeepers
PRO
335
57k
Transcript
吳䒭⠓爡ؒ؎ث٦ي DZNBך㼪Ⰵ✲⢽ ։"NB[PO1FSTPOBMJ[Fعٝؤؔٝ٥إىش٦։
h"UFBN *OD 荈䊹稱➜ 吳䒭⠓爡ؒ؎ث٦يٓ؎ؿأة؎ٕ 㛁遤䕵㆞ 71P& 菹馉 姻㸩
!LPQVH 䎃חؒ؎ث٦ي 䪮遭涪鿇חⰅ爡կ ך䖓ٕٓ٦ٝ✲噟ծؒٝة٦ذ؎ًٝٝز✲噟劤鿇ךًر؍،؟٦ؽأ✲噟 穗גծ䎃חؒ؎ث٦يٓ؎ؿأة؎ٕך㛁遤䕵㆞$50ח㽟⟣կ 䎃חؒ؎ث٦يٓ؎ؿأة؎ٕך 71PG&OHJOFFSJOHח㽟⟣׃ծ植㖈כؒ ؎ث٦ي &$✲噟劤鿇ך䪮遭顑⟣罏ה׃גDZNBח+0*/կ
h"UFBN *OD 湡如 ✲噟ך稱➜ "NB[PO1FSTPOBMJ[F鼅椚歋
㼪Ⰵ䧭卓 תה
h"UFBN *OD 湡如 ✲噟ך稱➜ "NB[PO1FSTPOBMJ[F鼅椚歋
㼪Ⰵ䧭卓 תה
㣐せ〢㾊ؽٕجؚٝ ㅷ䊛؎٦أزٙٝةٙ٦ 哅歊ꢻ䚈ؽٕ 秵♷⽆㢳⚥㣛ؽٕ ⠓爡嚊銲 爡せ 吳䒭⠓爡ؒ؎ث٦ي 鏣甧 䎃剢傈
劤爡䨽㖈㖑 䠥濼源せ〢㾊䋐 㣐せ〢㾊ؽٕجؚٝ' 䬿挿 匌❨٥㣐ꢻ٥状䀤 吳䒭Ⱅ䋐㜥 匌鏾♧鿇 爡㆞侧 せ䎃剢傈植㖈 㡰♳넝 ⭙ⰼ䎃剢劍 鸬穠㶨⠓爡侧 爡㕂ⰻ爡ծ嵲㢩爡
و٦؛حز 荈⹛鮦 ًر؍ٕؕ٥قٕأ؛، ♶⹛欵 ٓ؎ؿ؎كٝز ꆃ輐 ؒٝة٦ذ؎ًٝٝز 嶊顤頿 ➂勞
ؒ؎ث٦يכ؎ٝة٦طحز鯥ח✲噟遤ֲ筨さ*5⟰噟דׅ *5 طحز䪮遭٥؟٦ؽأ ًر؍، 䎢デ ؝ٝذٝخ *P5 &$ "* فٓحزؿؓ٦ي "3٥73 ٥ ٥ ٥ ٥ ٥ ٥
ؒ؎ث٦يך䲿⣘؟٦ؽأ
ؒ؎ث٦يך䲿⣘؟٦ؽأ
h"UFBN *OD չ؟؎وպך׀稱➜ 荈鯄鮦㼔 鸐顋؟؎ز 《䪔ְ荈鯄鮦珏겲⟃♳
㼔㾩侭⪒㡦ָ侭⪒ծּׅח⛦朐䡾ד荈㸓תדꂁ鷏 傈劤剑㣐秷ך荈鯄鮦㼔鸐顋؟؎ز 僴䎃䏝ך㡰♳넝כ秈⭙ⰼ չDZNB؟؎وպ
h"UFBN *OD չ؟؎وպך׀稱➜ ⴱךٖ؝ًٝر٦ءّٝؒٝآٝך㼪Ⰵ ٖ؝ًٝسכ䩛⹛ד鏣㹀׃גְ 㼪Ⰵ׃גְזַ椚歋כٖ؝ًٝسך⸬卓ָ⳿ח ְֻ䚍颵䭯גְ ず儗ח顋㡰ׅ荈鯄鮦ָ秈ㅷ湡ד&$ה׃גכ㼰זְ
곁㹏ךⱄ鏝㉏桦ָ⡚ֻծ곁㹏㾩䚍ָ飑Ⰵתדַזְ 㼪Ⰵ׃ג㹋ꥷך⸬卓然ְַ
h"UFBN *OD 湡如 ✲噟ך稱➜ "NB[PO1FSTPOBMJ[F鼅椚歋
㼪Ⰵ䧭卓 תה
h"UFBN *OD "NB[PO1FSTPOBMJ[F鼅椚歋 ⟃♴ךة؎فך؟٦ؽأ嫰鯰 "NB[PO1FSTPOBMJ[F ➭爡ךٖ؝ًٝر٦ءّٝؒٝآٝ • 剢겘顤欽ָ♰ⰼ䖓ך醡ㅷ侧珏겲
鐰⣣鯥כ如ךא 堣腉 剢겘顤欽 㼪Ⰵ䊨侧 麊欽؟ه٦ز ˟ٖ؝ًٝسך⸬卓כ⢪גזְהַזְךד鐰⣣鯥 ַכꤐ㢩
h"UFBN *OD Amazon Personalize • 5D;539'API"B4> C9
0 . • $&"+B4>C9 0 #". • "%>DA,=7 6? &+. • :382<1=@"+. • JavaScript"B4>C9 0 .+' • B4>C9 0. -/!. • )''*'+' • :382<1=@"(% 堣腉ך嫰鯰
h"UFBN *OD Amazon Personalize • 1&/,.0-$"* 1/,.0-
2& # *!.+0 • (8 &%* • ) $' ! • (1215& %* 剢겘顤欽ך嫰鯰
h"UFBN *OD 剢겘顤欽ך鑐皾麓玎 铬ꆃ㼎韋כչر٦ةך《鴥պծչزٖ٦صؚٝ 儗պծչٖ؝ًٝس䚍腉պךא ֿךֲչٖ؝ًٝس䚍腉պך嫰桦ָ㣐ְֹ 婍ך갪湡ךさ鎘כ ⰼ玎䏝ה䟝㹀
걼籕חزٖ٦صؚٝ׃זֻג葺ְהⴻ倖׃ ٖ؝ًٝس䚍腉כչؾ٦ؙ儗ך猱ךٖ؝ًٝ س㔐侧儗 儗傈 պד皾⳿ ؾ٦ؙ儗ך儗֮ך17ַ猱֮ךٖ؝ ًٝس㔐侧כ剑㣐⟝ה䟝㹀 ͑♰ⰼ ⰼ
h"UFBN *OD Amazon Personalize • !1.5)*30
+ • %(&'$ # # • !0.51) *1020+ • %(&'$ " 㼪Ⰵ䊨侧ך嫰鯰 չӛպה׃椚歋כ䊨侧ח㣐ֹז䊴ָזְ 鋅琎כ"NB[PO1FSTPOBMJ[Fךקֲָ侧٠剢⽃⡘ד䊨 侧ַַָה罋ִגְ
h"UFBN *OD 䟝㹀׃㼪Ⰵ䊨侧ךⰻ鏬 "NB[PO1FSTPOBMJ[Fךⴱ劍鏣㹀 傈 ٖ؝ًٝسٌرٕך⡲䧭
傈 ؟؎ز⩎ך堣腉ך㹋鄲 傈 ذأز٥رغحؚ 傈 さ鎘 傈
h"UFBN *OD 䟝㹀׃㼪Ⰵ䊨侧ךⰻ鏬 "NB[PO1FSTPOBMJ[Fךⴱ劍鏣㹀 傈 ٖ؝ًٝسٌرٕך⡲䧭
傈 傈ָر٦ةך彊⪒דծ傈ָٌرٕךزٖ٦صؚٝ ٌرٕךزٖ٦صؚٝכ"VUP.-ח⟣ׇ 堣唒㷕统ך濼陎כ䗳銲זְ ؟؎ز⩎ך堣腉ך㹋鄲 傈 ذأز٥رغحؚ 傈 さ鎘 傈
h"UFBN *OD 㹋ꥷך㼪Ⰵ䊨侧ךⰻ鏬 "NB[PO1FSTPOBMJ[Fךⴱ劍鏣㹀 傈 ٖ؝ًٝسٌرٕך⡲䧭
傈 傈 ؟؎ز⩎ך堣腉ך㹋鄲 傈 ذأز٥رغحؚ 傈 さ鎘 傈 䊨侧㟓⸇ך⚺ז⾱㔓כر٦ةך䭯倯ך锃侭ח儗ַַָ ֿהדׅկ ֿך⟝ろ㼪Ⰵך鑧כ2JJUBדչ"NB[PO1FSTPOBMJ[F 㼪Ⰵ׃גַךֿהպהְֲ鎸✲剅ֹת׃կ
h"UFBN *OD 㹋ꥷך㼪Ⰵ䊨侧ךⰻ鏬
h"UFBN *OD Amazon Personalize • 6,370!" $&&
• 6,370!+4/+ (&" • 6,370! !)01 *. ,7-5 -8 2.#'&% • 6,370!+4/+ ( & 麊欽؟ه٦زך嫰鯰 ٖ؝ًٝسך何㊣כٖ؝ًٝسך⳿׃倯ך锃侭זו "NB[PO1FSTPOBMJ[Fכٖ؝ًٝس䞔㜠ך䲿⣘ך
h"UFBN *OD 鐰⣣穠卓ךתה Amazon Personalize
h"UFBN *OD "NB[PO1FSTPOBMJ[F鼅椚歋 ♴鎸ך椚歋ַ"NB[PO1FSTPOBMJ[Fח寸㹀 㼛勻ٖ؝ًٝسךⵃ欽眔㔲䭁䓸׃װְׅ 8FC؟؎ز⟃㢩ד⢪ִ ٓٝصؚٝ؝أزָ㸜ְ 㼪Ⰵ䊨侧חקו㣐ֹז麩ְכזְ
麊欽؟ه٦زָזְֿהכ鏩㺁ׅ 㹋籐然钠ׅ➬穈כⴽד欽䠐ׅ
h"UFBN *OD 㼪Ⰵ䖓ך؟؎ز؎ً٦آ
h"UFBN *OD 湡如 ✲噟ך稱➜ "NB[PO1FSTPOBMJ[F鼅椚歋
㼪Ⰵ䧭卓 תה
h"UFBN *OD 湡如 ✲噟ך稱➜ "NB[PO1FSTPOBMJ[F鼅椚歋
㼪Ⰵ䧭卓 תה
h"UFBN *OD ⟃♴ךػة٦ٝ嫰鯰 "NB[PO1FSTPOBMJ[Fך)3//.&5"%"5" 遤⹛㾶娖ח㛇בְגٖ؝ًٝسׅ ِ٦ؠ٦׀הח穠卓ָ㢌 ػ٦اشٓ؎ؤٖ؝ًٝس "NB[PO1FSTPOBMJ[Fך4*.4
㼎韋ך荈鯄鮦ה⡂荈鯄鮦ٖ؝ًٝسׅ 铩ָ鋅גずׄ穠卓חז 䩛⹛ד鏣㹀 㼎韋ך荈鯄鮦ה⡂荈鯄鮦➂ך䩛ד鏣㹀ׅ 铩ָ鋅גずׄ穠卓חז
h"UFBN *OD زٖ٦صؚٝח欽ְ䞔㜠 ㉀ㅷ䞔㜠秈⟝ 顋㡰ָ穄✪׃㉀ㅷろדְ ㉀ㅷךꠘ鋮㾶娖湫鵚٠剢ⴓ秈♰⟝ 䎃ⴓ٥٠剢ⴓ٥ ٠剢ⴓךػة٦ٝדزٖ٦صؚٝ׃
הֿծ٠剢ⴓךػة٦ٝך礵䏝ָ葺ַ ⠓㆞䞔㜠כ⢪זְ )3//.&5"%"5"כ"VUP.-⢪ֲ "VUP.-⢪ֲה剑黝זػًٓ٦ة荈⹛ד鼅䫛ׅ 4*.4ךػًٓ٦ةכⴱ劍⦼⢪ֲ ➙ךהֿծ4*.4ד"VUP.-⢪欽דֹזְ
h"UFBN *OD ؙٔحؙ桦ה؝ٝغ٦آّٝ桦ך嫰鯰 14.97 1.43 7.62 0.84 11.55
1.03 0 2 4 6 8 10 12 14 16 HRNN-METADATA SIMS
h"UFBN *OD ٖ؝ًٝسٌرٕךًزؙٔأך嫰鯰 0.5559 0.3725 0.3665 0.2306 0.5149
0.3432 0 0.1 0.2 0.3 0.4 0.5 0.6 Normalized discounted cumulative Mean reciprocal rank HRNN-METADATA SIMS HRNN-METADATAAuto ML
h"UFBN *OD 㼪Ⰵ䧭卓ח㼎ׅ罋㻊 "NB[PO1FSTPOBMJ[Fכػ٦اشٓ؎ؤٖ؝ًٝس ⢪קֲָ⸬卓ָ⳿װְׅ )3//.&5"%"5"ד葺ְ穠卓ָ⳿גְ 4*.4ך穠卓כ)3//.&5"%"5"ה嫰ץה鋅⸋ׅ "VUP.-⢪ִל瀉儗ד葺ְ穠卓䖤
⢪זְ㜥さכ葺ְ穠卓ָ䖤תדػًٓ٦ةך锃侭 粸鵤ׅ䗳銲ָ֮ 锷椚甧גגػًٓ٦ة锃侭ׅחכ堣唒㷕统ך濼陎 ָ䗳銲
h"UFBN *OD 湡如 ✲噟ך稱➜ "NB[PO1FSTPOBMJ[F鼅椚歋
㼪Ⰵ䧭卓 תה
h"UFBN *OD תה "NB[PO1FSTPOBMJ[Fך㼪Ⰵכꨇ׃ֻזְ 堣唒㷕统ך濼陎כ♶銲 "NB[PO1FSTPOBMJ[Fכػ٦اشٓ؎ؤٖ؝ًٝس ⢪קֲָ⸬卓ָ⳿װְׅ "NB[PO1FSTPOBMJ[Fכ؟؎وך״ֲזٖ؝ًٝس
ח♶ⵃז䚍颵䭯؟؎زד⸬卓ָ֮ ず儗ח顋㡰ׅ荈鯄鮦ָ秈ㅷ湡ד&$ה׃גכ㼰זְ 곁㹏ךⱄ鏝㉏桦ָ⡚ֻծ곁㹏㾩䚍ָ飑Ⰵתדַזְ
h"UFBN *OD デ濼 *ODSFNFOUT➿邌嵲ꅿהؒ؎ث٦يؚٕ٦فぐ爡涪 顑⟣罏ךעׯֽ5BML/JHIU 剢傈 剢 ։!幪靼
ؒٝآص،ぢֽز٦ؙ؎كٝز❛崧⠓⪵׃ת ׅկ ˑؒ؎ث٦ي 涪顑⟣罏ز٦ؙ˒ד嗚稊 5FDIQMBZ PS$POOQBTT ֶַ歍鴥ְֻׁ
չזד䎊ׇחז⠓爡חֿׅהպ չ➙ַ䎃竲ֻ⠓爡חֿׅהպ