- Главная
- Разное
- Дизайн
- Бизнес и предпринимательство
- Аналитика
- Образование
- Развлечения
- Красота и здоровье
- Финансы
- Государство
- Путешествия
- Спорт
- Недвижимость
- Армия
- Графика
- Культурология
- Еда и кулинария
- Лингвистика
- Английский язык
- Астрономия
- Алгебра
- Биология
- География
- Детские презентации
- Информатика
- История
- Литература
- Маркетинг
- Математика
- Медицина
- Менеджмент
- Музыка
- МХК
- Немецкий язык
- ОБЖ
- Обществознание
- Окружающий мир
- Педагогика
- Русский язык
- Технология
- Физика
- Философия
- Химия
- Шаблоны, картинки для презентаций
- Экология
- Экономика
- Юриспруденция
Spotify’s Music Recommendations Lambda Architecture презентация
Содержание
- 1. Spotify’s Music Recommendations Lambda Architecture
- 2. Overview Why Lambda? Use Case: Discover Recommendations Batch Architecture Real-time Architecture Challenges Future Work
- 3. Why Lambda? 1 new user every 3
- 4. Discover Recs
- 5. The Discover Page Algorithmically generated fresh recs for users.
- 6. The Discover Batch Pipeline
- 7. Machine Learning Deep Dive
- 8. Word2Vec Words with similar contexts have similar meaning
- 9. Word2Vec King – Man + Woman = Queen
- 10. Annoy Approximate Nearest Neighbors Oh Yeah! https://github.com/spotify/annoy
- 11. Batch Architecture Strengths Recs based on complete
- 12. Intro to Storm
- 13. Storm Distributed real-time computation system
- 14. Storm @ Spotify
- 15. Real-time Architecture
- 16. Workers die -> Cascading JVM Process death
- 17. Future/Ongoing Work Simplify the topology Keep listens
- 18. Questions Esh Kumar eshvk@spotify.com Emily Samuels esamuels@spotify.com
Overview Why Lambda? Use Case: Discover Recommendations Batch Architecture Real-time Architecture Challenges Future Work
Слайд 2Overview
Why Lambda?
Use Case: Discover Recommendations
Batch Architecture
Real-time Architecture
Challenges
Future Work
Слайд 11Batch Architecture
Strengths
Recs based on complete user history
Weakness
User vector generation time increasing
with no. users.
Not reflective of current mood.
Not reflective of current mood.
Слайд 16Workers die -> Cascading JVM Process death
Memcache flakiness
Cassandra JVM problems due
to write/overwrite pattern
Challenges
Слайд 17Future/Ongoing Work
Simplify the topology
Keep listens for 24 hours
Ongoing work on other
real time personalization features.
