NetDraw

NetDraw is a free Windows program for visualizing social network data, written by Steve Borgatti (of Analytic Technologies). It ships bundled with, and is installed automatically alongside, the commercial UCINET analysis package (UCINET), though it can also be run standalone. Within the social-network-analysis (SNA) toolchain, NetDraw occupies the drawing-and-layout niche: UCINET computes metrics on matrices, while NetDraw renders the resulting graphs, letting an analyst move nodes by hand, encode attributes visually, and export publication figures.

Its data model is a graph of nodes (actors) and ties (relations) that explicitly supports 1-mode and 2-mode (bipartite) networks, multiple relation types (multiplex data) held in one file, and both binary and valued ties. Node attributes—categorical or quantitative, and either external (inherent properties) or internal (structural measures)—can be mapped to color, shape, and size, while tie color and line thickness encode relation type or strength. Valued ties can be dichotomized interactively, stepping through cut-off thresholds to show only stronger or weaker links.

NetDraw reads UCINET system files, UCINET DL text files (UCINET DL), Pajek datasets (Pajek), and its own native VNA text format (Netdraw VNA). VNA is a plain-text format organized into asterisk-prefixed "star" sections such as Node data, Node properties, and *Tie data, each with a column header row; it stores attribute values (including text) plus display state—node coordinates, colors, sizes, and visibility—so a saved diagram reopens exactly as left. Because VNA is sparse-friendly, NetDraw handles fairly large graphs (thousands to ~10,000 nodes with sufficient RAM). Export options include Pajek and UCINET formats plus raster/vector graphics (Windows Metafile, BMP, JPEG, GIF). VNA is widely interchanged: importers exist in tools like Gephi (Gephi), and Pajek (Pajek) shares closely related conventions.

Layout options include random, circular, attribute-coordinate, MDS, and spring-embedding placement, and an Analysis menu adds structural tools: components, isolates, k-cores, cutpoints, cliques, factions, Newman–Girvan community detection, and ego-network extraction, often fed back into visual encodings (e.g. sizing nodes by betweenness). Limitations are those of node-link drawing generally: very large or dense multiplex networks become visual "hairballs" that reveal little, the tool is Windows-only, and it targets manual exploratory figure-making rather than scripted, reproducible pipelines like those built with NetworkX or igraph.