Open source diagramming framework for Java
|Raw data is hard to read||Project and browse|
|Common approaches to graph layout fail to deliver representation where we can actually read things, either globally, or locally.||Data projections lead to readable data diagrams at any scale. Projected data makes layout tuning easy and flexible.|
|Global and local views without compromise|
Zooming through nested structures is great to let the user discover explore the various layers of complexity of a structure.
Triggering local views on individual items open the door to in-depth data exploration.
|Compose data diagrams with multiple layouts in a single map|
Assigning layouts to structural data patterns generates diagrams closed to the domain model conventions.
2 ways for analysing data topologies:
|Multiple layout algorithms working together|
|Layouts are designed to be adaptive and to deal with constraints given by the user or by other concurrent layouts.|
|Label layouts||Slot layouts||Edge layouts||Edge bundling|
|Text is made always readable by a background job and remains reponsive to the visualization context (zoom change, positions change, etc).||Highly interconnected entities can be explored in depth with precise control over slot layouts.||Edges can be drawn with various policies: bundling edges to simplify the view, using an orthogonal path, bending edge with splines [not yet available], etc.||Edge bundling summarizes information and allows generalized knowledge about groups relationships. Detailed neighbourhood can be explored by dynamically by browsing to different views, or by adding context aware information according to currently displayed area [not yet available], etc).|
|Layouts can be mixed freely. For example, one may build orthogonal bundled edges by applying specific layout for all incoming and outgoing edges (interfaces) of each node.|
|Nested layouts reduce computational cost|
Resolving several small layouts is easier than resolving one large layout.
Domain model oriented layout are often straightforward to compute (single step instead of multi-step such as force based layouts).
|Composite layouts are ideal candidates for parallel computing|
Running on the Java Virtual Machine, Datagr4m’s runtime can use various concurrent computation policies to make best use of mono and multi-core architectures.
Datagr4m can render dense structure efficiently by relying on the GPU.
Hierarchical layouts are more stable than flat graph layouts: a classical force based layout might become instable by adding new nodes, whereas projection maintain map consistency over time.