Output more agent & edge information

[lkettle]

Right now, Hashkat reports agent, followers and their status as tweeters or retweeters. So no information is output relating to agent (node) language, region or ideology, and no information is output as to tweets (edge) language, tweet type, etc.

So the scenarios that can be run and analyzed and visualized are limited.

For example, I want to run a simulation with two differerent ideology regions redland & blueland,
speaking different languages, french & english, with varying degrees of uni-linguality, bilinguality, to see how ideologic tweets spread.

And it can't be done. The information is there (in the agent struct) it just needs to be reported.

[itamblyn]

@lkettle Can you please research what the best file format is to write this out? Some of these quantities are time dependent. Can our current output format support that? What do the common viz tools look for with time dependent graphs?

[lkettle]

Points to Ponder

1. a simple data dump can be used by ANY visualization program (csv, text)
2. our choice will limit the user to our choice.
3. our choice will become stale quickly (note Gephi in all our visualizations is already outdated)
4. right now you have placeholder output files for gexf, graphml, and a third.

If we give an option for big data dumps in simple form, (ie all agent info, all edge info, plus same at a couple of time intervals), we can use whatever visualizations we want to demonstrate the program, I may like Gephi, Nick may like NetworkX, someone else may like Cytoscape etc. and the user similarly has that option.

[itamblyn]

I mean what are the advantages of the existing formats. I think a lot of
tools take lots of different formats

On Fri, Jan 15, 2016, 11:59 AM lkettle notifications@github.com wrote:

Points to Ponder

1. a simple data dump can be used by ANY visualization program (csv, text)
2. our choice will limit the user to our choice.
3. our choice will become stale quickly (note Gephi in all our
   visualizations is already outdated)
4. right now you have placeholder output files for gexf, graphml, and a
   third.

If we give an option for big data dumps in simple form, (ie all agent
info, all edge info, plus same at a couple of time intervals), we can use
whatever visualizations we want to demonstrate the program, I may like
Gephi, Nick may like NetworkX, someone else may like Cytoscape etc. and the
user similarly has that option.

—
Reply to this email directly or view it on GitHub
#94 (comment).