Graph Analytics

Graph algorithms for centrality, community detection, pathfinding, and similarity, callable as SQL and Cypher.

Centrality

• ArticleRankCompute ArticleRank centrality, a PageRank variant tuned for citation-style graphs
• Betweenness CentralityCompute betweenness centrality measuring how often a node lies on shortest paths
• Closeness CentralityCompute closeness centrality measuring how close a node is to all others
• Degree CentralityCompute in-degree, out-degree, and total degree for all nodes
• Eigenvector CentralityCompute eigenvector centrality via power iteration on the adjacency matrix
• Harmonic CentralityCompute harmonic centrality, the sum of reciprocal shortest-path distances
• HITS (Hubs and Authorities)Compute mutually reinforcing hub and authority scores for every node
• PageRankCompute PageRank centrality scores for all nodes in a graph

Community Detection

• ConductanceCompute per-community conductance for a fixed node-to-community partition
• Connected ComponentsFind connected components (groups of mutually reachable nodes)
• K-Core DecompositionAssign each node its coreness number, the largest k for which the node belongs to a k-core
• Label PropagationDetect communities by iterative neighbor-label voting until labels stabilize
• Leiden Community DetectionDetect well-connected communities by combining Louvain local moves with a refinement phase
• Local Clustering CoefficientCompute the per-node local clustering coefficient and triangle count
• Louvain Community DetectionDetect communities using the Louvain modularity optimization algorithm
• ModularityCompute the modularity score Q for a fixed node-to-community partition
• Strongly Connected ComponentsFind strongly connected components in a directed graph
• Triangle CountCount the number of triangles each node participates in

Embeddings

• FastRP Node EmbeddingsCompute Fast Random Projection node embeddings via iterated sparse projections of the row-normalised adjacency matrix
• Random Walk Corpus GenerationGenerate a corpus of fixed-length random walks starting from every node in the graph

Pathfinding

• A* Shortest PathHeuristic-guided single-target shortest path search between two nodes
• Bellman-Ford Single-Source Shortest PathsSingle-source shortest paths that admit negative edge weights and detect negative cycles
• Breadth-First SearchTraverse the graph in breadth-first order from a source node
• Delta-Stepping Single-Source Shortest PathsParallel single-source shortest paths via bucketed light/heavy edge relaxation
• Depth-First SearchTraverse the graph in depth-first order from a source node
• Minimum Spanning TreeCompute the minimum spanning tree of a weighted graph
• Shortest PathFind the shortest weighted path between two nodes
• Yen's K-Shortest PathsEnumerate the K shortest loopless paths between two nodes in increasing weight order

Query Language

• CALLInvokes a procedure with an optional per-CALL device hint and YIELD for result columns.
• CALL { subquery }Executes an inner query as a subquery.
• Execution HintsControl execution device (CPU/GPU) and property loading mode (streaming/eager).
• LIMIT (Graph)Limits the number of returned results.
• MATCHPattern matching clause -- finds nodes and relationships in the graph.
• Node PatternMatches a node with optional variable, labels, and properties.
• OPTIONAL MATCHLike MATCH, but returns NULLs for missing parts instead of filtering rows.
• ORDER BYSorts results by one or more expressions.
• Relationship PatternMatches a relationship with direction, type, and optional variable-length path.
• RETURNSpecifies the output columns and expressions.
• shortestPath / allShortestPathsFinds shortest path(s) between two nodes.
• SKIPSkips the first N results (pagination).
• UNIONCombines results from multiple queries.
• UNWINDExpands a list into individual rows.
• WHEREFilters results based on a predicate expression.
• WITHPipes results between query parts, enabling aggregation and filtering.

Similarity

• K-Nearest NeighborsFind the k most similar nodes for each node based on graph structure
• Node SimilarityCompute pairwise similarity between nodes based on shared neighbors

Topology

• Articulation Points (Cut Vertices)Identify vertices whose removal disconnects a connected component of the graph
• Bridge EdgesEnumerate every edge whose removal increases the number of connected components