#KNEKTR 
Features
Functional Integration
Allows for the easy integration of custom functionality to KNEKTR instances.
var coalesce = function (){ /* ... */ }
var knektr = new KNEKTR();
/* integrate function with KNEKTR */
knektr.integrate(coalesce);
/* call from knektr */
knektr.coalesce();
var Qmaximization = function (){ /* ... */}
var graph = knektr.addGraph('myGraph');
/* integrate function with a single graph instance */
graph.integrate(Qmaximization);
/* call from graph */
graph.Qmaximization();
var graphTaxonomy = function (){ /* ... */}
var graphs = knektr.graphs;
/* integrate function with a graph collection */
graphs.integrate(graphTaxonomy);
/* call from graphs */
graphs.graphTaxonomy();Goals
This is similar to some functionality provided in Sigma.js, though the option of scoping is made available in KNEKTR as well as a simpler interface.
D3.js shortcomings: While D3 is a fantastic general purpose library for data visualization in the browser, it does not provide a great deal of flexibility and power for creating, visualizing, and manipulating graphs specifically.
The D3 interface for visualizing nodes and edges is entirely through a force directed layout object. this force object accepts a list of node objects and a list of edge (link) objects. The source and target properties which define a link do not accept nodeIDs, rather the location of nodes within the supplied node list. This is highly prone to error when dealing with dynamically changing graph environments where nodes are continuously being inserted and deleted.
In general, D3 does not provide a graph model layer, but simply a visualization layer. This means that D3 does not natively support simple graph analysis or querying. Because of this, D3 does not let you natively operate on additional node or edge properties (e.g. node/edge filtering, selective traversals, etc.). Wouldn't it be nice to multiply realize your backend data model with your front end API?
VivaGraph shortcomings: While VivaGraph supplies an intuitive and modular approach to a graph library (separation of graph model from rendered visual), the ease-of-use remains a large draw-back. Now, VivaGraph has been modularized into distinct ngraph modules, which all must be imported for certain functionality to work. Also, the documentation is not very good.