[Air-L] Large Twitter network graph

[kalev leetaru]

Depending on the structure of the graph, it is also worth considering the
sfdp engine in GraphViz for the visualization portion - while not as good
at teasing out fine structure as other algorithms, it is one of the more
scalable layout and software rendering algorithms for large graphs and
we've used it for some very large Twitter graphs. Back in 2012 we routinely
performed layout and rendering of billion-edge graphs in under 30 minutes
on a single core on an SGI UV2 and GraphViz makes it quite straightforward
to enrich the graphs with external calculated attributes that can be used
to affect the visuals. Today we still make heavy use of its rasterizer for
geographic visualization for rendering high-resolution graphs containing
tens to hundreds of billions of points and it is still one of the more
efficient software rasterizers out there and have continued to use it for
our largest social graph layout needs, while using Gephi for small graphs.

Kalev



On Mon, Sep 7, 2020 at 9:42 PM Oluwole Badmus <obadmus at uwo.ca> wrote:

> Hi Samuel,
>
> Depending on the bus speed/processing capacity of your computer,
> visualizing large samples can be (obviously) problematic. However, some viz
> software (such as Gephi, UCNet, Netlyzer, etc.) can be tweaked (settings)
> to accommodate large node mass visualization. I assumes your best bet is to
> use these software solely designed for visualization, it helps with
> computing resource management (a list of useful tools has been compiled by
> SocialMedia Lab at Ryerson https://socialmedialab.ca/apps/).
>
> Similarly, check out Deen Freelon’s TSM tools on GitHub ...in case you’re
> working with Twitter data. It comes in well recommended for preparing node
> tables for Gephi parsing.
>
> HTH
>
> Best wishes
> Martins Olu’ Badmus
> LIS PhD Candidate
> Western University
> Rm 4039 FNB
> London, Ontario. Canada N6A 5B9
> Skype: olubadmus007
>
> I acknowledge that Western University is located on the traditional lands
> of the Anishinaabek, Haudenosaunee, Lūnaapéewak and Attawandaron peoples,
> on lands connected with the London Township and Sombra Treaties of 1796 and
> the Dish with One Spoon Covenant Wampum. This land continues to be home to
> diverse Indigenous peoples (e.g. First Nations, Métis and Inuit) whom we
> recognize as contemporary stewards of the land and vital contributors of
> our society.
>
> From: Olaniran Psalmuel<mailto:psalmuel35 at gmail.com>
> Sent: September 7, 2020 5:14 PM
> To: air-l at listserv.aoir.org<mailto:air-l at listserv.aoir.org>
> Subject: [Air-L] Large Twitter network graph
>
> Hello all,
>
> I am in need of help and suggestions on a task for my study. I have a
> dataset with over 1 million nodes that I am trying to graph. My interest is
> in visualising the degree of connections, centrality and betweenness. Used
> Networkx module in python but the code for drawing the graph runs for days
> with no error or output. I have tried out a smaller sample of the dataset
> and it worked nicely.
> Any suggestions on a tool/software that can handle such node size, or an
> alternative approach would be greatly appreciated.
> Thank you in advance.
>
> Samuel Olaniran
> PhD Student,
> Department of Media Studies,
> University of the Witwatersrand
> _______________________________________________
> The Air-L at listserv.aoir.org mailing list
> is provided by the Association of Internet Researchers http://aoir.org
> Subscribe, change options or unsubscribe at:
> http://listserv.aoir.org/listinfo.cgi/air-l-aoir.org
>
> Join the Association of Internet Researchers:
> http://www.aoir.org/
>
> _______________________________________________
> The Air-L at listserv.aoir.org mailing list
> is provided by the Association of Internet Researchers http://aoir.org
> Subscribe, change options or unsubscribe at:
> http://listserv.aoir.org/listinfo.cgi/air-l-aoir.org
>
> Join the Association of Internet Researchers:
> http://www.aoir.org/