A Visual Bibliography of Tree Visualization | Hacker News Reader

A Visual Bibliography of Tree Visualization

treevis.net

153 pointsescot10 comments

That one

https://treevis.net/#Li2020a

or that one

https://treevis.net/#Brath2012

look useful for seeing 6 million simultaneous messages in exchange, the salient content peculiar to you you rewire as strands in zero gravity depending on emphasis and the A.I. assistant has AST extrapolations at focal points

I don't think anything will top the squarified treemap for me, but I love this page.

I really don't get the appeal of the sunburst style visualizations. They seem like the worst possible use of area to [fail to] convey information.

I think I agree with arboles that a tree visualization wouldn't be appropriate for this page. However, this reminds me of an old idea I had, to present tree-based data structures in a tree. I was thinking something like a phylogenetic tree representing the evolution of these data structures over time. Too much of a research project for me though.

Regarding arguments made for "sunburst style visualizations", it is usually that in Sunburst visualizations the drawing space increases with each level as the circumference of the radial plot increases. This is useful for trees, which usually have more nodes to show on lower levels closer to the leaves, than on higher levels closer to the root.

In comparison to the favored squarified treemap layout, sunburst visualizations are also able to represent an order in which the nodes may be given.

Regarding a tree visualization for the contents, we tried this in the beginning, but it often needs a lot of "editorializing" to decide which class to put a visualization technique in, if it would fit 2 or even 3 classes at the same time. Hence for the web-version, I switched over to a plain list. The old hierarchical layout can still be found at https://cs.au.dk/~hjschulz/pdfs/treevispo.pdf

I was hoping the related techniques would themselves be organized as a tree! But doesn’t seem like it.

Of course this begs the question why they aren't represented by some sort of tree visualization.

  Tree visualization is one of the best-studied areas of information visualization; researchers have developed more than 200 visualization and layout techniques for trees. The treevis.net project aims to provide a hand-curated bibliographical reference to this ever-growing wealth of techniques. It offers a visual overview that users can filter to a desired subset along the design criteria of dimensionality, edge representation, and node alignment. Details, including links to the original publications, can be brought up on demand. Treevis.net has become a community effort, with researchers sending in preprints of their tree visualization techniques to be published or pointing out additional information.

-- Site itself is a paper (https://ieeexplore.ieee.org/document/6056510)

Very interesting. I especially like the filtering and the older examples. Too bad most are behind paywalls.

Edit: Somebody else had the same thought heh.

Bunch of other types of visualization collections in the dropdown top-left. Appears to be suffering from link rot.

OP any backstory?

Has there been a study of how well any of these work in VR ? I'd like to know which ones are intuitive and comfortable for navigating with hand motions and viewer teleportation.

Unlike the other comments, I don't think the site about tree organizations should be organized like a tree. The authors are well read in tree visualizations, in the site you can filter by "Dimensionality" or "Representation" or "Alignment", and you can combine these tags. The authors don't use trees because they know the limitations.

This reply isn't filed under the intended two posts it should reference because of limitations of how comments are structured on HN.

to the limitations on how comments are structured, as trees, not DAGs, and not just on HN but nearly everywhere: have you ever encountered a "forum" or "chat" eher you can kind of star merge several threads with a single reply?

>>33567235

I can only think of Futaba-style discussion boards, that is, 4chan. On 4chan threads are linear, but you're supposed to reply to posts by mentioning their unique post nº. For example, the above comment would want to mention >>33566411 and >>33566389. You can mention as many posts as want. The posts get a backlink, so the thread is a navigable, true graph structure. Reading discussion on 4chan can be confusing as a deep chain of replies can branch out to mention a recent surface post, and to understand what's going on you have to both delve down focused reply chains but also be aware of incoming posts that are displayed cutting across any one discussion.

Here is a typical 4chan thread that demos this:

https://desuarchive.org/g/thread/87053939/#87081210

I chose to link a post where a picture visualizes a 4chan thread in a directed graph. Also, a screen above there's a "Thanks." to several other posts, in one post.

Yes. One of the biggest ironies is that the very best approach to open forum discussion is applied to some of the least useful uses of the application. The UI is as slapped together and confronting as the conversations but the proper branching graph system could make everything from search to navigation to LLM AI digestion of huge messaging systems just work better.

Reply on news.ycombinator.com