GUESS

GUESS (the Graph Exploration System) is an interactive tool for the visualization and exploratory analysis of graphs and networks, created by Eytan Adar, largely developed while he was at HP Labs and continued in his academic work (later at the University of Michigan). Its defining idea, introduced in the 2006 CHI paper "GUESS: A Language and Interface for Graph Exploration," is the tight coupling of a graphical front end with an embedded scripting language, so that objects typed in an interactive interpreter are bound directly to the nodes and edges being drawn on screen. This lets analysts rapidly prototype visualizations and run ad hoc queries against a live graph rather than through a batch pipeline.

The scripting layer is Gython, a domain-specific extension of Python implemented on top of Jython (Python on the JVM). Gython adds graph-specific operators, primitives, and syntactic sugar for selecting node/edge sets, filtering by attribute, styling visual properties, and invoking layout and analysis routines; because it runs on the JVM it integrates with Java graph libraries, notably JUNG (the Java Universal Network/Graph Framework), for algorithms and data structures. Being a full programming language rather than a fixed menu of commands is GUESS's main differentiator from click-only visualization tools.

GUESS's native interchange format is GDF (Graph Data Format), a simple CSV-like plain-text format split into a nodedef section and an edgedef section, each led by a typed header line followed by comma-separated rows. GDF outlived GUESS itself: it is still recognized as format:gdf and is a documented importer in tool:gephi. Beyond GDF, GUESS imports format:pajek and format:gml files and reads format:graphml, and it exports static images in raster and vector formats (PNG, JPG, GIF, EPS, PDF, SVG).

In the graph-data and format-conversion landscape GUESS sits alongside other network-visualization environments such as tool:gephi, tool:cytoscape, tool:pajek, and tool:socnetv, and its scripting-first philosophy anticipates library-driven analysis in tool:networkx and tool:igraph. Its notable limitations are age and maintenance: the software stalled at a 1.0.x beta in the mid-2000s, depends on an aging Java/Jython stack, targets moderate rather than very large graphs, and is effectively unmaintained today, so most of its practical legacy survives through the widely supported GDF format rather than the application itself.