Skip to main content

Demo Datasets

The examples use four public datasets. MerRec is the largest: eight questions over 1.27 billion marketplace events, from the brands that moved toward Barbie's audience to the candidate lists for a Pokemon campaign. The other three pages apply the same pattern to a citation graph, a product catalog and a malware benchmark. Each section states a question, shows the SQL and its saved results, and explains what the composition supports.

DatasetDomainSize used by the examplesExamples
MerRecMercari US marketplace events, May through October 20231.268B events, 82.6M items, 5.6M usersBarbie's July, shopping worlds and their bridges, a campaign list from graph, vectors and a model, lookalikes, a budgeted model, catalog duplicates, and a session-key check
Citation NetworkDBLP / AMiner V12 papers, April 2020 snapshot4.9M papers, 45.6M citation edgesResearch communities, a ForceAtlas2 neighborhood, shared references, and citation paths
Amazon ReviewsAmazon Reviews 2023, Electronics43.9M reviews, 1.6M products; 1,024-dim BGE-M3 vectors for 1M products and 1M reviewsTravel-headphone shortlist, changing review groups, and original review text
EMBER2024Malware benchmark feature vectors1.56M Win32 training files, 359,994 test files; 6,315 challenge filesClassifier errors, model contributions, and shared import patterns