Hi, Yes writing a dynamic GEXF file from the logs you are describing would work. The GEXF dynamic support work like that: for each node or edges you can define one or several time intervals. Intervals can be dates or numbers. Better support and interaction with dynamics is one of the top project's priorities.The GEFX Explorer doesn't support dynamics for the moment, although it's in the developper's roadmap.
Yes I think we can do that, it would be great to have a Neo4j-Gephi SoC proposal. Thanks for pointing Neoeclipse, it's summing up your needs in a sense. I'm thinking how to integrates such features in Gephi. We have a modular software architecture, built on top of Netbeans Platform. It's easy to extend existing features and add plug-ins. So I think it's not a problem to port existing Neoeclipse features to Gephi plugins and nicely maintain it after. But could you please explain me how Neoeclipse reads and updates the Neo4j instance in a few words, or pointing some resources? For us, we would like to see a robust and fast interaction layer between Neo4j instance and our Graph API. That would fulfill our off-memory graph structure aim in the roadmap [1]. Loading very large graphs in Gephi, even if it is scalable will always make the memory suffer at one moment. Having the complete Graph in Neo4j, plus the ability to quickly get a sub-graph with a query and save updates would be a great achievement :-) Cheers, Mathieu [1] http://wiki.gephi.org/index.php/Roadmap On Tue, Mar 23, 2010 at 6:03 PM, Tobias Ivarsson < tobias.ivars...@neotechnology.com> wrote: > Hi! > > I've had a look at Gephi and it looks really interesting. One particular > thing I found interesting is the support for "dynamic graphs". I'm hoping to > be able to use that for creating animated sequences of how a traversal moves > through a graph. Not only would this be cool for demonstrations, but it > would also be very useful for visual debugging of applications. I've written > a small framework that captures all actions in a Neo4j Graph Database and > forwards them to a logging system, my thinking is that this logging system > would create a GEXF[1] file that can then be reviewed in Gephi or GEXF > Explorer[2]. Do you think that would work? Or would it be abusing the GEXF > format? > > Peter said that he talked to you over IRC earlier, and that you are a > mentoring organization for Google Summer of Code. If you guys would find > that interesting we would be very happy to see a SoC project about combining > Neo4j and Gephi. I could write up a suggestion for such a project for your > wiki, and volunteer to as a mentor. I have previous experience with SoC as a > student 2007, 2008 and 2009, Google Summer of Code is so far the only way > I've managed to get payed to work on Jython. Among the projects that have > already been proposed I would like to put a vote on "Dynamic attributes and > statistics"[3] since that could tie in nicely with the visual debugging > system I outlined above. But being able to work with a Neo4j database from > within Gephi is a much cooler project ;) > > We have a visualization toolkit for Neo4j that is similar to what I would > like to see from an integration between Neo4j and Gephi, but not nearly as > advanced and visually pleasing as Gephi. It's called Neoclipse[4] and is an > extension to Eclipse. My hope is to have Gephi be able to do all the things > that Neoclipse can do and more (and especially prettier), at the end of the > Summer of Code project. > > Cheers, > Tobias > > [1] http://gexf.net/ > [2] http://gexf.net/explorer/ - I don't know yet if this supports dynamics > though (http://gexf.net/format/dynamics.html) > [3] > http://wiki.gephi.org/index.php/Google_Summer_Of_Code_2010#Dynamic_attributes_and_statistics > [4] http://wiki.neo4j.org/content/Neoclipse > <http://wiki.neo4j.org/content/Neoclipse> > On Tue, Mar 23, 2010 at 12:00 PM, Mathieu Bastian < > mathieu.bast...@gmail.com> wrote: > >> Hi, I'm Gephi main developper. >> >> I think Gephi fits to your needs because it can handle very large graphs, >> and propose innovative layout algorithms. On our 8gb ram machine, we >> handle >> 300K nodes and 1.5M edges, rendering is slow but it's still fine. >> >> But the problem you're describing is a layout issue. With traditionnal >> Spring repulsion or Fruchterman-Reingold algorithm, you have a O(N²) >> algorithm. Scaled to millions of nodes and edges, computing require hours >> for each algorithm pass. In Gephi we recently implement Yifan Hu's layout >> algorithm, that has a 0(nlog(n)) complexity. I recommed you to try this >> out. >> This problem is often references as multilevel graph layout algorithm. >> >> Cheers >> _______________________________________________ >> Neo mailing list >> User@lists.neo4j.org >> https://lists.neo4j.org/mailman/listinfo/user >> > > > > -- > Tobias Ivarsson <tobias.ivars...@neotechnology.com> > Hacker, Neo Technology > www.neotechnology.com > Cellphone: +46 706 534857 > _______________________________________________ Neo mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
