We added a new feature that allows you to browse your oink network as a graph:
As I wrote before in this blog, Oinker’s document oriented data is not well suitable for graph representation (nodes and edges) because an oink network tends to have nodes with large amount of text or other miscellaneous data that would be meaningless or too verbose as a node in the graph display.
To keep a graph reasonable to browse, we have to find some way to simplify an oink network. I’ve been thinking about this for some time and recently I came up with an idea of “topic nodes“.
Oinker’s graph visualization doesn’t display all the nodes connected to your room’s board. Instead it displays the nodes selected by a certain rule and we call them topic nodes. How are topic nodes selected? Though the way of selection is subject to change, currently a topic node is:
- a node whose content has only one word or sentence.
- a node whose content length is shorter than or equal to 30 (in English. It becomes shorter than that in multibyte character languages).
- a node whose content is not Markdown (starting with
- a node whose content is not a URL
- a node whose content is not a file
With this topic nodes visualization, Oinker now has a good balance between abstract and detail level. To begin with, Oinker adopted the document oriented style because we believe it should be essential to collect and accumulate detailed information (a.k.a contexts) as much as you can when you want to find valuable ideas or concepts. Mind maps are great to share an overview of an idea. However, without the context, most of its value will be lost in interpretations by each individual.
In Oinker, the graph view is one of the ways to overview the results of your knowledge-building. It’s like an observation platform on a mountain, which allows you to view the panorama after you climbed the mountain.
One thought on “Oink Network Graph Visualization”
I think that the decision to use short names as a selection scheme is similar to the decision made at http://debategraph.org/ to include a short label in addition to a long label, in addition to a details (hidden from graph view) section. That way, each node can have a graph display with a mouse-over with deeper explanation. If you choose that node, it opens very much like a blog post.