@unrdf/graph-analytics

Advanced graph analytics for RDF knowledge graphs using graphlib.

Features

1. RDF to Graph Conversion

Convert RDF triple stores to graphlib Graph instances for analysis.

import { createStore } from '@unrdf/oxigraph';
import { rdfToGraph, getGraphStats } from '@unrdf/graph-analytics';

const store = createStore();
// ... add RDF triples ...

const graph = rdfToGraph(store);
const stats = getGraphStats(graph);
console.log(`Nodes: ${stats.nodeCount}, Edges: ${stats.edgeCount}`);

2. Centrality Algorithms

Compute entity importance using PageRank and degree centrality.

import { computePageRank, getTopNodes } from '@unrdf/graph-analytics';

const pagerank = computePageRank(graph, { dampingFactor: 0.85 });
const top10 = getTopNodes(pagerank, 10);
console.log('Most important entities:', top10);

3. Path Finding

Discover relationships through shortest paths and path enumeration.

import { findShortestPath, findAllPaths } from '@unrdf/graph-analytics';

const shortestPath = findShortestPath(graph, sourceUri, targetUri);
console.log('Path:', shortestPath.path);

const allPaths = findAllPaths(graph, sourceUri, targetUri, { maxDepth: 4 });
console.log(`Found ${allPaths.length} paths`);

4. Community Detection

Identify clusters and communities in knowledge graphs.

import { detectCommunitiesLPA, getCommunityStats } from '@unrdf/graph-analytics';

const communities = detectCommunitiesLPA(graph);
const stats = getCommunityStats(communities);
console.log(`Detected ${stats.totalCommunities} communities`);

Installation

pnpm add @unrdf/graph-analytics

Quick Start

# Run demo
pnpm demo

# Run tests
pnpm test

# Run with coverage
pnpm test -- --coverage

API Reference

Converter

• rdfToGraph(store, options) - Convert RDF store to graph
• sparqlResultsToGraph(results, options) - Convert SPARQL results to graph
• getGraphStats(graph) - Compute graph statistics

Centrality

• computePageRank(graph, options) - PageRank algorithm
• computeDegreeCentrality(graph) - Degree centrality
• computeBetweennessCentrality(graph) - Betweenness centrality
• getTopNodes(scores, k) - Get top K nodes by score

Paths

• findShortestPath(graph, source, target) - Dijkstra shortest path
• findAllPaths(graph, source, target, options) - All paths enumeration
• findCommonNeighbors(graph, node1, node2) - Common neighbors
• findKHopNeighbors(graph, source, k) - K-hop neighborhood
• discoverRelationshipChains(graph, source, target) - Relationship patterns

Clustering

• detectCommunitiesLPA(graph, options) - Label propagation
• detectCommunitiesModularity(graph, options) - Modularity-based
• findKCore(graph, k) - K-core decomposition
• getCommunityStats(communities) - Community statistics

Performance

Optimized for RDF knowledge graphs:

• Efficient graph conversion (streaming support)
• Fast PageRank convergence
• Scalable path algorithms
• Memory-efficient community detection

License

MIT