Query and analyze
A useful first query answers a small question about a graph you understand. Inspect its card, classes and example queries before writing your own: the names used in RDF are exact identifiers, so a plausible name can still match nothing.
Find an answer
Use the Playground to query without installing anything, or choose a language client. The SPARQL reference describes supported syntax and limitations; the CLI reference covers query commands and output formats.
For a specific kind of question:
- GeoSPARQL filters geometry and time.
- Federation combines graphs; it explains the differences
between a union, a cross-file join and a remote
SERVICEcall. - Reasoning uses ontology relationships to derive answers that are not written explicitly in the data.
- Semantic zoom gives an overview of classes and communities before you inspect individual entities.
Check the data and the cost
SHACL validation checks a graph against shapes: for example, whether each entity has a required property. Read the reported violations alongside the shape that produced them.
Range reads avoid loading a whole file at open time. They do not make every query small: a broad scan, join or large result set can still read and retain substantial data. Start with a selective question and inspect its cost; see the CLI reference and compatibility and limits.
Compute beyond SPARQL
Graph algorithms explains how to project an edge list, compute with Python libraries, and write results back to a new graph. Topic modeling labels communities by their text; multi-criteria communities combines structural and attribute-based partitions. These are different analyses, so choose by the question you want to answer rather than by the visualization.