Data Science Education. Building Knowledge Graphs

on YouTube ↗41m

" summary

Builds a public index that currently contains 12,000 resources (mostly videos) by scraping MOOC platforms, YouTube and lecture sites, extracting transcripts and slides, mapping metadata to Schema.org + JSON-LD, and storing it in ElasticSearch. Uses random-forest classifiers (F1 ≈ 0.8) to filter high-quality videos, LDA topic modeling for visualization, and Wikipedia clickstream signals to infer prerequisites; human curators correct tags and feed models. Links people and organizations to DBpedia/DBLP/ORCID linking and releases the dataset under an academic noncommercial license.

AI-written summary · about

" chapters (25)

1. 0:00Introduction to talk
2. 2:33Project overview and team
3. 5:43BD2K program and training resources
4. 7:13Identifying training materials
5. 9:32YouTube and ELIXIR data sources
6. 11:01Building the resource index
7. 12:24Resource database and contributors
8. 13:50Web scraping and metadata extraction
9. 14:59YouTube quality classification
10. 16:44Mapping to schema.org
11. 17:46Public dataset and licensing
12. 18:31Resource representation and RDF
13. 19:14Automatic tagging with ontology
14. 20:39Curator interface and feedback
15. 22:47Linking to knowledge graph
16. 24:11Entity resolution with DBpedia
17. 27:25Evaluation of linking accuracy
18. 28:28Enhanced resource descriptions
19. 30:19Resource dependencies and prerequisites
20. 31:42Wikipedia clickstream analysis
21. 34:08Knowledge graph features
22. 34:52Topic modeling and visualization
23. 36:17Future work and improvements
24. 37:22Personalization and recommendations
25. 39:40Q&A session

" related talks

• 30mHow Graph Data Science Can Turbocharge Your Knowledge GraphKristof Neys · Big Data LDN 2022
• 30mFast and scalable domain-specific knowledge graphs generationOmar Alonso · Scale By The Bay 2019
• 42mBuilding a knowledge graph for climate policyHarrison Pim, Fred O'Loughlin · Berlin Buzzwords 2025
• 29mBuilding a knowledge graph with the Natural History Museum to work as a taxonomic equivalence engineAndy Symonds, Siobhan Hughes · XConf Europe 2023
• 31mConnecting Keywords to Knowledge Base Using Search Keywords and WikidataFang Xu · PyData Berlin 2016
• 30mMaking Your Knowledge Graph LLM-Ready: Quality Assessment and Design Strategies for GraphRAGPanos Alexopoulos · NODES 2025
• 18mFree Texts to Knowledge Graphs - Holy Grail of Semantic Computing on an Industrial ScaleBernd Geiger · Big-Data.AI Summit 2022
• 37mCorinna Cortes, Head of Research, GoogleCorinna Cortes · MLconf NYC 2015