Like Kei, I'm very suspect of the value 3D brings if the relationships themselves are more or less "2D" or dimensionless, and the number of data elements (and relationships) is relatively large. For that matter, even 2D representations of the total graph space are not all that communicative. What it can bring is a really busy UI that chews up a lot of GPU and CPU cycles, and can generate some cool wall art...
When applied to a meaningful and manageable subset of nodes, it can provide some utility. In terms of 3D, since the network of nodes is effectively "shapeless", you'd need to define/associate/specify some type of dimensionality to make it valuable - e.g. the X/Y plane represents people and group relationships and the Z dimension represents products they buy... I think more work needs to be done in thinking about the "what and why" in advance of the "how". -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Christopher Schmidt Sent: Tuesday, October 12, 2010 2:07 PM To: Neo4j user discussions Subject: [SPAM] Re: [Neo4j] Graph visualization Kei, you mean the answer (...point, ...insight) is mostly 'no'? I hope that it depends on the type of data you want to show. Although it is an issue to create layouts in a 3D space that adds some visual value (we are only used to use 2D presentations). Maybe the possibility to rotate, fly through and live highlight (by changing colors etc.) of nodes and assocs can bring this value. Christopher On Tuesday, October 12, 2010, Keiichiro Ono <[email protected]> wrote: > Hi. > Except small and sparse graph, 3D force-directed/spring model > automatic layouts creates big hairball and it is very hard to browse. > I've tried igraph and it creates nice 3D visualizations, but in many > cases, our users (mostly scientists) say, "it's cool...but what's the > point? Does it give us new insights by extra dimension?" > > Are there any people working on this fundamental problem? > > Thanks, > Kei > > 2010/10/12 Alex Averbuch <[email protected]>: >> Hey, >> igraph already supports 3D layouts and makes the vertex coordinates >> programmatically accessible. Maybe Jung or something similar (and Java >> based) would offer the same functionality? >> >> igraph is written in Python so not the idea solution, but for a proof of >> concept you could hardcode the coodinates that igraph gives you. Then once >> the actual visualizations are working, work on your own (or integrate with >> another library) layout algorithms. >> >> On Tue, Oct 12, 2010 at 11:34 AM, Christopher Schmidt < >> [email protected]> wrote: >> >>> I think that the most tricky thing will be the algorithm, that places the >>> nodes and associations in a 3D space. >>> >>> Christopher >>> >>> Am 2010 10 12 11:08 schrieb "Andreas Kollegger" < >>> [email protected]>: >>> That would be super cool. 3D could be beautiful, and possibly allow more >>> interesting visualizations of a graph. In addition to an overview of the >>> "scene", it would be fun to play with 1st person and 3rd person views of >>> the >>> current node. >>> >>> What would be an easy proof-of-concept? >>> >>> /Andreas >>> >>> >>> On Oct 12, 2010, at 1:19 AM, Andrew Andkjar wrote: >>> >>> > I have not seen one in my Internet travels, h... >>> _______________________________________________ >>> Neo4j mailing list >>> [email protected] >>> https://lists.neo4j.org/mailman/listinfo/user >>> >> _______________________________________________ >> Neo4j mailing list >> [email protected] >> https://lists.neo4j.org/mailman/listinfo/user >> > > > > -- > Keiichiro Ono http://www.keiono.net/ > _______________________________________________ > Neo4j mailing list > [email protected] > https://lists.neo4j.org/mailman/listinfo/user > -- Christopher twitter: @fakod blog: http://blog.fakod.eu _______________________________________________ Neo4j mailing list [email protected] https://lists.neo4j.org/mailman/listinfo/user _______________________________________________ Neo4j mailing list [email protected] https://lists.neo4j.org/mailman/listinfo/user
