Knowledge Graphs & Deep Learning at YouTube
summary
Describes combining knowledge graphs with deep learning at YouTube to extract a video's main topics from audiovisual content, title/description text, and contextual signals. Explains using those inferred topics to enhance recommendations, search, structured browsing, and ad targeting.
AI-written summary · about
chapters (20)
- 0:00Intro
- 0:40Generic Recs
- 1:22Personnalized Recs
- 2:31Collaborative Filtering
- 3:20Deep Learning for Recommendations
- 4:50Cold Start Problem
- 5:17Rescued by Entities
- 6:41Visual Content Analysis (VCA)
- 10:37Wikipedia: Knowledge for Humans
- 11:01Wikidata: Knowledge for Machines
- 11:19Wikidata: Statements
- 11:53Wikidata: Graph of Entities
- 12:12Wikidata: SPARQL
- 13:21Wikidata: Query using Python
- 13:34Search: Knowledge Panels
- 14:37Structured Discovery
- 16:47Entity Linking using Textual Metadata
- 16:58Entity Recognition Using Deep Learning
- 18:18Entity Linking Using Contextual Signals
- 18:31Ranking Candidate Entities
related talks
38mKnowledge Graphs : Adding a New Dimension to Vector Search and Generative AI
05mAugmenting Visual Information in Knowledge Graphs for Recommendations
40mKnowledge graphs --enter- the Hype Cycle
29mKnowledge Graph–Driven Feedback Loops to Optimize TikTok Shop Video Marketing
30mHow Graph Data Science Can Turbocharge Your Knowledge Graph
22mThe Nexus of Innovation: How Knowledge Graphs Fuel AI Scientific Breakthroughs
44mBuilding Knowledgeable Machines
23mML-Powered Search & Recommendation 101