SNAP Package Description

• SNAP Directory Structure
• Example Applications
• Functionality of the Core SNAP Library

SNAP Directory Structure

The SNAP package contains the following directories:

• snap-core:the core SNAP graph library;
• snap-adv:advanced SNAP components, not in the core, but used by examples;
• snap-exp:experimental SNAP components, still in development;
• examples:small sample applications that demonstrate SNAP functionality;
• tutorials:simple programs, demonstrating use of various classes;
• glib-core: STL-like library that implements basic data structures, like vectors (TVec), hash-tables (THash) and strings (TStr), provides serialization and so on;
• test:unit tests for various classes;
• doxygen:SNAP reference manuals.

Example Applications

• agmfit: Detects network communities from a given network by fitting the Affiliation Graph Model (AGM) to the given network by maximum likelihood estimation.
• agmgen: Implements the Affiliation Graph Model (AGM). AGM generates a realistic looking graph from the community affiliation of the nodes.
• bigclam: Formulates community detection problems into non-negative matrix factorization and discovers community membership factors of nodes.
• cascadegen: Identifies cascades in a list of events.
• cascades: Simulates a SI (susceptible-infected) model on a network and computes structural properties of cascades.
• centrality: Computes node centrality measures for a graph: closeness, eigen, degree, betweenness, page rank, hubs and authorities.
• cesna: Implements a large scale overlapping community detection method for networks with node attributes based on Communities from Edge Structure and Node Attributes (CESNA).
• circles: Implements a method for identifying users social circles.
• cliques: Finds overlapping dense groups of nodes in networks, based on the Clique Percolation Method.
• coda: Implements a large scale overlapping community detection method based on Communities through Directed Affiliations (CoDA), which handles directed as well as undirected networks. The method is able to find 2-mode communities where the member nodes form a bipartite connectivity structure.
• community: Implements network community detection algorithms: Girvan-Newman, Clauset-Newman-Moore and Infomap.
• concomp: Computes weakly, strongly and biconnected connected components, articulation points and bridge edges of a graph.
• flows: Computes the maximum network flow in a network.
• forestfire: Generates graphs using the Forest Fire model.
• graphgen: Generates undirected graphs using one of the many SNAP graph generators.
• graphhash: Demonstrates the use of TGHash graph hash table, useful for counting frequencies of small subgraphs or information cascades.
• infopath: Implements stochastic algorithm for dynamic network inference from cascade data, see Structure and Dynamics of Information Pathways in On-Line Media
• kcores: Computes the k-core decomposition of the network and plots the number of nodes in a k-core of a graph as a function of k.
• kronem: Estimates Kronecker graph parameter matrix using EM algorithm.
• kronfit: Estimates Kronecker graph parameter matrix.
• krongen: Generates Kronecker graphs.
• lshtest: Implements locality sensitive hashing.
• magfit: Estimates Multiplicative Attribute Graph (MAG) model parameter.
• maggen: Generates Multiplicative Attribute Graphs (MAG).
• mkdatasets: Demonstrates how to load different kinds of networks in various network formats and how to compute various statistics of the network.
• motifcluster: Implements a spectral method for motif-based clustering.
• motifs: Counts the number of occurence of every possible subgraph on K nodes in the network.
• ncpplot: Plots the Network Community Profile (NCP).
• netevol: Computes properties of an evolving network, like evolution of diameter, densification power law, degree distribution, etc.
• netinf: Implements netinf algorithm for network inference from cascade data, see Inferring Networks of Diffusion and Influence
• netstat: Computes structural properties of a static network, like degree distribution, hop plot, clustering coefficient, distribution of sizes of connected components, spectral properties of graph adjacency matrix, etc.
• randwalk: Computes Personalized PageRank between pairs of nodes.
• rolx: Implements the rolx algorithm for analysing the structural roles in the graph.
• testgraph: Demonstrates some of the basic SNAP functionality.
• zygote: Demonstrates how to use SNAP with the Zygote library, which significantly speeds up computations that need to process the same large graph many times.

Functionality of the Core SNAP Library

Brief description of SNAP functionality implemented in various files in the distribution package: