Unlocking graph analytics in DuckDB with SQL/PGQ
summary
Creates a SQL/PGQ property-graph layer on top of per-label tables and translates graph patterns into relational plans so DuckDB executes them natively. Claims duckpgq yields 10–100x better analytic performance than Neo4j on benchmark queries, implements specialized UDFs and a compact graph data structure for BFS-style path finding (handles any shortest paths), and exposes algorithms as table functions like PageRank, weakly connected components and local clustering coefficient. Installable as a community extension.
AI-written summary · about
related talks
45mGraph databases, PostgreSQL and SQL/PGQ
27mExploring property graphs with SQL/PGQ in PostgreSQL
24mpg_duckdb: Ducking awesome analytics in Postgres
31mWhen property graphs join SQL
41mGraphGen: Conducting Graph Analytics over Relational Databases
21mWhat can Postgres learn from DuckDB?
29mAll about graphs in Database
49mModern Graph Databases