← back
A Practitioner's Guide to Graphs - Tim Ainge, Good Collective
Takeaway
Graph value comes from consistent modeling and algorithms that exploit relationships, not from graph visualization alone.
Summary
- Structured graph extraction improves when agents receive a domain schema instead of an unconstrained subject-predicate-object template.
- Ontology guidance standardizes entity names and units, while entity resolution prevents duplicate nodes from fragmenting relationships.
- Embeddings can support flexible entity matching, combining AI extraction with graph structure.
- Cypher traversal and personalized PageRank reveal multi-hop relationships and relevance; examples include recipe discovery, Pinterest recommendations, and legal citation chains.
knowledge-graphsentity-resolutionpagerank
Original description
A speed-run through the basics... what is a graph, extracting graphs from unstructured text, schema first and ontological improvements and then a slightly more detailed discussion of personalised page rank, shortest path and subgraph matching algorithms. Every idea is explored with an explanation of the principles, and accessible example with code and linked to real-world implementations or references to articulate the value and relevance. Speakers: Tim Ainge (Good Collective): Software Engineering practitioner in startups, private and public sector organisations. X/Twitter: https://x.com/timainge LinkedIn: / timainge GitHub: https://github.com/timainge/