Algorithm Availability and Dependencies

The following table lists several key characteristics of each algorithm,

Besides these requirements, each algorithm is designed to work with certain types of graph data, such as directed edges vs. undirected edges. Those data-driven requirements are listed with each algorithm.

Maturity - The maturity of an algorithm is one of three classifications:

  • Alpha = basic functionality and testing

  • Beta = full-featured, well-tested

  • Production = full-featured, stable, rigorously tested, optimized for speed and resource efficiency, suitable for production use

GraphStudio - Availability for easy installation as one of GraphStudio’s preloaded algorithms.

Packaged Template - Availability in TigerGraph’s built-in package of templated algorithms with just-in-time compilation.

UDF Required - This algorithm requires the installation of a custom user-defined function (UDF). If installed by GraphStudio or as a packaged template query, the UDF will be taken care of automatically.

For security reasons, the ability to install a UDF is disabled by default. Your system administrator can enable this capability if appropriate.

Subquery Required - This algorithm requires the installation of a custom subquery.

Category Algorithm Maturity Graph Studio Packaged Template UDF Required Subquery Required

Centrality

ArticleRank

Prod

Y

Y

Centrality

Betweenness Centrality

Prod

Y

Y

Centrality

Closeness Centrality

Prod

Y

Y

Centrality

Closeness Centrality, approximate

Prod

Y

Centrality

Degree Centrality

Prod

Y

Y

Centrality

Degree Centrality, weighted

Prod

Y

Y

Centrality

Eigenvector Centrality

Prod

Y

Centrality

Harmonic Centrality

Prod

Y

Centrality

Influence Maximization, CELF

Prod

Y

Centrality

Influence Maximization, greedy

Prod

Y

Centrality

PageRank

Prod

Y

Y

Centrality

PageRank, weighted

Prod

Y

Y

Centrality

Personalized PageRank

Prod

Classification

Greedy Graph Coloring

Prod

Y

Y

Classification

k-Nearest Neighbors (Cross-Validation Version)

Prod

Y

Classification

k-Nearest Neighbors (Batch Version)

Prod

Y

Classification

k-Nearest Neighbors (Cross-Validation Version)

Prod

Y

Y

Classification

Maximal Independent Set, deterministic

Prod

Y

Y

Community

k-Core Decomposition

Beta

Y

Community

Label Propagation

Beta

Y

Y

Community

Label Propogation, speaker-listener

Beta

Y

Y

Community

Local Clustering Coefficient

Beta

Y

Y

Community

Louvain

Beta

Y

Y

Community

Strongly Connected Components

Prod

Y

Community

SCC (Small-World Optimized)

Prod

Y

Y

Community

Triangle Counting

Prod

Y

Community

Triangle Counting, fast

Prod

Y

Y

Community

Weakly Connected Components

Prod

Y

Y

Community

WCC (Small-World Optimized)

Prod

Y

Graph ML

Fast Random Projection

Alpha

Y

Y

Graph ML

Weisfeiler-Lehman Isomorphism

Prod

Y

Y

Y

Path

A*

Beta

Y

Path

Breadth-First Search

Beta

Y

Y

Path

Cycle Detection

Beta

Y

Y

Path

Estimated Diameter

Beta

Y

Path

Maximum Flow

Beta

Y

Path

Minimum Spanning Forest

Beta

Y

Path

Minimum Spanning Tree

Beta

Path

Single-source Shortest Path (Unweighted)

Prod

Y

Path

Single-source Shortest Path (Weighted)

Prod

Similarity

Cosine Similarity of Neighborhoods (All Pairs, Batch)

Prod

Y

Y

Similarity

Cosine Similarity of Neighborhoods (Single-Source)

Prod

Y

Similarity

Jaccard Similarity of Neighborhoods (All Pairs, Batch)

Prod

Y

Similarity

Jaccard Similarity of Neighborhoods (Single Source)

Prod

Y

Similarity

Vector Similarity

Beta

N/A

is a UDF

Topological Link Prediction

Adamic Adar

Prod

3.9.2

Y

Topological Link Prediction

Common Neighbors

Prod

3.9.2

Y

Topological Link Prediction

Preferential Attachment

Prod

3.9.2

Y

Topological Link Prediction

Resource Allocation

Prod

3.9.2

Y

Topological Link Prediction

Same Community

Prod

Y

Y

Topological Link Prediction

Total Neighbors

Prod

3.9.2

Y