How Big Data Makes Content More Relevant in Search
summary
Describes a production pipeline that clusters search behavior into meaning not keywords: a graph-based Topic Explorer built over almost a billion queries that groups intents (nutrition vs. growing trees for “avocado”), turns hundreds of noisy keyword suggestions (700+ from AnswerThePublic or Keyword Planner) into manageable clusters, and scores pages with a content score to predict relevance (example: an image-gallery page scores ~17%). Claims three and a half years of algorithm and data‑science work and machine‑learning vectors to map topics to answers.
AI-written summary · about
chapters (7)
- 0:00Content marketing challenges
- 0:51Data-driven content strategy
- 2:54Understanding user intent
- 4:42Topic Explorer tool
- 7:45Keyword vs topic concepts
- 11:15Using questions for content
- 12:27Measuring content success
related talks
39mRelevant search results (with elasticsearch)
46mFrom Naive Sorting to Natural Language Processing, 3 Years of Optimization of Our Search Engine
30mSearch on Steroids — A Story of Semantics
31mWhen humans teach machines
45mIntegrate your Search Engine (...)
20mLayering every technique in RAG, one query at a time
26mThe Future of Search
57mDiscover semantic topics in a collection of documents