Building a knowledge graph for climate policy
summary
An open-source climate policy knowledge graph maps ~1,500 curated concepts across 2M pages and 45M passages and is already powering a search used by ~350K users/year. It stores concepts in Wikibase, generates one-to-one classifiers from a factory (80 trained, ~74 live) and runs inference into Vespa with Prefect orchestration and Pulumi IaC. Indexing drops from 137 days to under a day. Classifiers are human-evaluated with stratified samples, translations, and a qualitative vibe-checker.
AI-written summary · about
chapters (8)
- 0:00Introduction and overview
- 2:08Why build a knowledge graph
- 7:25Concepts and relationships
- 12:17Building classifiers
- 19:47Evaluating classifiers
- 20:08Scaling the pipeline
- 32:12Results and future plans
- 34:00Q&A session
related talks
41mData Science Education. Building Knowledge Graphs
29mBuilding a knowledge graph with the Natural History Museum to work as a taxonomic equivalence engine
44mDemocratizing Climate Science
30mHow Graph Data Science Can Turbocharge Your Knowledge Graph
30mFast and scalable domain-specific knowledge graphs generation
36mThe OpenClimate Network: Nested and Connected Climate Accounting for a Digitally...
26mBuilding an Entity-Resolved Knowledge Graph: Why and How
42mOS-Climate: A Breakthrough Open Data and Open Source Analytics Collaboration for...