NetAnalysys

NetAnalysys is a high-performance terminal-based graph analysis toolkit written in Rust. It loads graph datasets, computes structural metrics, estimates shortest-path distances via landmark methods, and provides interactive exploration — all within a TUI.

Features

• Multi-format graph loading — .txt, .csv, .mtx with automatic ID remapping
• Graph characterization — type, order, size, density, degree distribution
• Connectivity analysis — weak & strong (Tarjan's SCC) components, largest component fraction
• Diameter estimation — Double BFS, random-sampled BFS, snowball sampling
• Triangle counting & clustering — global and average clustering coefficients
• Robustness analysis — LCC degradation under random and hub-targeted vertex removal
• Landmark-based distance estimation — Random, Highest-degree, and Farthest-first selection strategies
• Accuracy benchmarking — landmark estimates vs exact BFS over random vertex pairs
• Visualization — terminal-printed degree distribution charts + PNG export via plotters
• Performance logging — structured duration logs for all pipeline stages

Datasets

The project has been validated on publicly available real-world graph datasets, located in graphs/:

Dataset	Type	Source
web-Google	Directed	Google web graph
web-NotreDame	Directed	Notre Dame web graph
web-Stanford	Directed	Stanford web graph
Wiki-Vote	Directed	Wikipedia voting network
soc-wiki-Vote	Directed	Wikipedia vote matrix (MTX)
CA-AstroPh	Undirected	ArXiv Astro Physics co-authorship
CA-GrQc	Undirected	ArXiv General Relativity co-authorship
ca-coauthors-dblp	Undirected	DBLP co-authorship
Email-EuAll	Undirected	Email network from EU institution
musae_git_edges	Undirected	GitHub developer social network
com-orkut.ungraph	Undirected	Orkut social network (large)
com-youtube.ungraph	Undirected	YouTube social network (large)
vk	Undirected	VK social network (large)

Quick Start

cargo run

Select a graph file from the built-in file browser, choose a diameter estimation method, and the analysis pipeline runs automatically.

Requirements

• Rust stable toolchain
• Terminal with crossterm support

Usage

1. Launch the application
2. Load a graph dataset (samples in graphs/)
3. Select diameter method: Double BFS, Random 500 BFS, Snowball sampling, or All
4. Review computed metrics, terminal charts, and saved PNG plots
5. Optionally enter the interactive landmark mode for pairwise distance queries and benchmarks

Supported Formats

Format	Separator	Comments
.txt	whitespace	Lines starting with # are ignored
.csv	comma	Leading/trailing whitespace trimmed
.mtx	whitespace	MatrixMarket: % comments, first non-comment line is the header

Graph direction is inferred from the parent folder name: Directed/ → directed, Undirected/ → undirected, default is undirected.

Architecture

src/
  analysis/     — Metrics: connectivity, degree, diameter, robustness, triangles, clustering
  graph/        — Graph representation (adjacency list, ID mapping) and BFS/DFS traversal
  landmarks/    — Landmark selection strategies (Random, Highest-degree, Farthest-first)
  interactive_landmarks/  — TUI for landmark-based distance queries and benchmarking
  parser/       — File format detection and parsing
  ui/           — Terminal UI (menus, degree chart rendering, PNG export)
  main.rs       — Orchestration pipeline and entry point

The graph uses compact internal vertex IDs with bidirectional mapping to external dataset IDs. Computations are parallelised with Rayon and Tokio where applicable.

Performance

Processing time scales with graph size. Small datasets (up to ~100K edges) complete in milliseconds. The full analysis pipeline on com-orkut (~117M edges) takes approximately 22 minutes on a standard office laptop without overclocking.

Output Artifacts

• degree_data1.png — degree distribution plot
• log_degree_data.png — log-log degree distribution plot
• performance.log — per-stage execution timings
• trace.log — pipeline trace events

Development

cargo build              # Compile
cargo test               # Run tests
cargo fmt                # Format code
cargo clippy --all-targets --all-features  # Lint

Contributing

Contributions are welcome. Please open an issue to discuss changes before submitting a pull request. Ensure the project builds without warnings and passes cargo clippy.

License

MIT — see LICENSE.