Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Sign up for free
Menu
Search
Features
All features
Private URLs
Password Protection
Custom URLS
Scheduled publishing
Remove Branding
Restrict embedding
Deck Collections
Notes
Features
All features
Private URLs
Password Protection
Custom URLS
Scheduled publishing
Remove Branding
Restrict embedding
Deck Collections
Notes
Explore
Featured decks
Featured speakers
Programming
Technology
Storyboards
Explore
Featured decks
Featured speakers
Programming
Technology
Storyboards
Pricing
Search
Sign in
Sign up for free
2026TECHFRESH畢業分享會 - Lightning Talk - E起 See S...
Search
Sponsored
·
SiteGround - Reliable hosting with speed, security, and support you can count on.
→
LINE Developers Taiwan
PRO
June 17, 2026
Technology
3.3k
0
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
2026TECHFRESH畢業分享會 - Lightning Talk - E起 See See : 電商推薦讀心術? 數據說了算
講者: Jenaya
活動:
https://techfresh.landpress.line.me/20260616/
LINE Developers Taiwan
PRO
June 17, 2026
More Decks by LINE Developers Taiwan
See All by LINE Developers Taiwan
Zona Hello World AIDD PPT
line_developers_tw
PRO
0
410
Evan 2026-09-12 國立臺中科技大學 x LINE PROTOSTAR 創新創業競賽- LINE 技術應用說明
line_developers_tw
PRO
0
390
Evan 2026-09-12 國立臺中科技大學 x LINE PROTOSTAR 創新創業競賽- LINE API應用介紹
line_developers_tw
PRO
0
360
Zona Codex Community Meetup 分享
line_developers_tw
PRO
0
1.4k
Evan & Zona COSCUP 2026 分享
line_developers_tw
PRO
0
370
台大資管營企業參訪-Bowen Chen
line_developers_tw
PRO
0
150
2026TECHFRESH畢業分享會 - AI 時代的人生存檔點
line_developers_tw
PRO
0
3.4k
2026TECHFRESH畢業分享會 - 原生還是跨平台? App 開發踩坑實錄
line_developers_tw
PRO
0
3.4k
2026TECHFRESH畢業分享會 - 葬送的通靈師:化系統與用戶雜訊成行動訊號
line_developers_tw
PRO
0
3.3k
Other Decks in Technology
See All in Technology
CSS2026:OWS研究会・招待講演「世界観を、実装する」
atsush1_sada
2
340
Claude for Google Workspaces ポン出し/claude-for-google-workspace-slide-conversion
emiki
0
13k
cdk deployとcdkd deploy、実際にデプロイして5倍速近くの差を実測した
yama3133
1
110
Don't write the architecture only in the prompt.
neohsu
0
170
CTO Night & Day Morning Session_AI-DLC:生成 AI 時代の開発プロセス変革
pikosan0000
0
120
AWSDirectConnectの構築の進め方を整理してみた
masa_my
1
200
書き手が人でもAIでも正しくなる道、SREを埋め込んだDeveloper Platform / tech drivers day8
chmikata
4
800
AI秘書と毎日働くのはもう当たり前。開発以外の業務に足りない機能をどう埋める?
minorun365
PRO
4
1.9k
Agent × Maestro で始める モバイル E2E 端末準備からテスト実行・結果確認まで
sansantech
PRO
0
230
Snowflake でできることが増えたら、 ABAC が必要になった
kaeton
0
110
渡米前の準備を制す!re:Invent 2026徹底準備術 / The Ultimate Preparation Guide for re:Invent 2026
yuj1osm
1
350
「Learning Outcome」を書きたくない私が、Learning Outcomeと折り合いをつけるまで
mhlyc
0
250
Featured
See All Featured
AI Search: Implications for SEO and How to Move Forward - #ShenzhenSEOConference
aleyda
1
1.4k
Designing Experiences People Love
moore
143
24k
End of SEO as We Know It (SMX Advanced Version)
ipullrank
3
4.5k
How GitHub (no longer) Works
holman
316
150k
Rebuilding a faster, lazier Slack
samanthasiow
85
9.7k
jQuery: Nuts, Bolts and Bling
dougneiner
66
8.6k
Scaling GitHub
holman
464
140k
Making the Leap to Tech Lead
cromwellryan
135
10k
The B2B funnel & how to create a winning content strategy
katarinadahlin
PRO
1
580
The Web Performance Landscape in 2024 [PerfNow 2024]
tammyeverts
12
1.3k
Neural Spatial Audio Processing for Sound Field Analysis and Control
skoyamalab
0
580
Exploring the relationship between traditional SERPs and Gen AI search
raygrieselhuber
PRO
4
4.4k
Transcript
E 起 See See: 電商推薦讀心術? 數據說了算! LINE EC Data Intern,
Jenaya Lin
01 02 03 推薦版位與推薦架構 推薦結果分析 團隊介紹 CONTENT
Jenaya Lin LINE TechFresh EC Data Team Master's Student, NCCU
MIS
職能介紹 Data Scientist 4 Data Engineer Machine Learning Engineer •
ML Theory, Algorithm • Data Analysis • Data Pipeline • Database • Platform • Model Implementation • Tuning • MLOps
可以在這邊 找到我們! EC 5
推薦 版位 商店推薦 6 商品推薦 文章推薦
Data Extraction Extract raw event logs from Hive tables 推薦架構
7 Feature Engineering Model Training Generation Preparing user-side behavioral features and item-side features Train embeddings with Two-Tower architecture Orchestrate DAG pipelines for personalized & trending inference
離線 v.s. 線上評估 8 • 環境:使用歷史數據 • 目的:在模型上線前,驗證模型對於已知用戶 行為的預測能力 •
優勢:迭代速度快、不會影響實際用戶 • 缺點:缺乏商業指標、與真實環境存在差距 • 環境:真實環境 • 目的:衡量推薦系統對業務指標(如點擊率 、轉換率)的實際貢獻 • 優點:反映真實商業價值、捕捉動態反饋 • 缺點:時間與工程成本高、具備業務風險、 易受外在因素干擾 離線評估 (Offline Evaluation) 線上評估 (Online Evaluation)
離線評估 線上與離線評估比較圖 - 檢驗兩者趨勢是否相同 1 2 除 5/7 外,線上與 離線評估趨勢一致
離線評估指標 9 公式 意義 Precision@10 前 10 筆命中數 / 10
推薦前 10 筆中「命中」的比例 Recall@10 前 10 筆命中數 / 該用戶 label 總數 所有喜好商品中,前 10 筆能抓到幾個 MAP AP = Σ(P@i × rel(i)) / |labels|, 再取用戶平均 考慮命中位置的加權 Precision,越靠前的命中 越加分 NDCG@10 DCG@10 / IDCG@10, DCG = Σ rel(i)/log₂(i+1) 位置折扣累積增益,命中越靠前分數越高,以 理想排序做歸一化 MRR 1/rank(第一個命中) 取用戶平均 第一個推薦命中的排名越前越好;只看「有沒 有猜到最重要的一個」
離線評估 商品推薦 - 七天內各項指標個人化與流行推薦 1 0 個人化推薦很成功/ 消費者的需求明確 且分散
離線評估 商店推薦 - 七天內各項指標個人化與流行推薦 1 1 個人化推薦仍有進步 空間/用戶話題集中
實習心得
LinkedIn: jenaya-lin