Visualization of Networks

This page is about an idea that could be implemented in CouchSurfing in the future. See also Future features and functionality.

Currently possible ways to visualize

• XFN and xfngraph, check Java web start

• http://graphviz.org/

• http://turrican.unixag-zw.fh-kl.de/studianalyse/ - social network visualization software , free - open source - maybe one of us wants to test it ? - related to Prefuse ;

English Presentation Pdf ;
Its Video

Future

Can be implemented ( ? ) in Couchsurfing integrating ( one of ) the following Open Source Free Softwares:

• http://prefuse.org/
• http://www.deepamehta.de/

They can be downloaded.

• http://sourceforge.net/project/downloading.php?groupname=prefuse&filename=prefuse-beta-20060715.zip&use_mirror=mesh
• http://www.deepamehta.de/install/client/login.html

They relate to my brainstorming on

• http://oikoumene.coforum.net/wikka.php?wakka=ReferenceMaps

Another interesting library is JSViz and it is entirely done in javascript so it can be integrated directly in the web site. It seems also be very light on the client side (I guess visualizing a network of 1 million node would kill the browser). I think it is Free Software, see http://kylescholz.com/projects/speaking/tae2006/music/#B00005RGIZ and http://www.kylescholz.com/blog/2006/06/force_directed_graphs_in_javas.html and http://kylescholz.com/projects/wordnet/ and how to build a direct graph from xml. The source is available under a Creative Commons Attribution License. See http://www.kylescholz.com/blog/2006/10/jsviz_031_available.html --Phauly 10:42, 11 January 2007 (EST)

http://del.icio.us/deliciousdante/visualization

http://del.icio.us/deliciousdante/ReferenceMaps

Visualization of Networks such as :

http://jheer.org/vizster/

http://www.cs.berkeley.edu/~jheer/vizster/vizster.wmv ( short film presentation )

http://standpedia.com/

Extend to broader Reference Maps?

http://oikoumene.coforum.net/wikka.php?wakka=ReferenceMaps

links to other tools and related texts:

http://del.icio.us/deliciousdante/referencemaps?setcount=100

Why

How

0. before deciding which type of visualisation tool at cs could be used for common decisions, brainstorming ... , a test drive could be performed with some of the most promising tools. for this test drive there could be a constructed problem, that these visualisation tools must be able to solve. the constructed problem could look like this:

• constructed problem for test-driving online visualisation-programms: programming new features for an online travel-exchange community.

there are 1 mio. members in that online travel-exchange community. you are an admin with programming skills. due to recent complaints about hosts who made their guests feel bad during their stay, you think about coding a new feature that blocks user's profiles which got five negative references. you want feedback for your idea in order to know what the "community" thinks on this.

the feedback needs to be concise, ideally maybe like a picture that you have one look at and think "okay, i do it" or " no, i don't do it" or "i gotta explain the idea a bit further". because of the fact that you have time to work on the code in about two weeks you would like to have the feedback on that feature you suggested (blocking user's profiles that got five negative references) in two weeks.

via the feedback users should have the possibility to suggest a feature B you could work on, in case the idea (blocking profiles) get's turned down. via the feedback the coder (you) should have the possibility to recognize tendencies and shades that make clear at one look how the idea you suggested should be implemented (e.g. most people want people's profiles with five negative references to be marked with a red x on the top left corner and not blocked) or that the question for feedback must be more precisely or which other problems should be considered before implementing that idea in order to solve the initial problem.

testing some visualisation tools

Here you can see how visualisation tools work for the question "Shall user-profiles with more than 5 negative comments be treated in special ways?". If you want you can add your or other points of view, but remember this is a test of the tool not necessarily a means to come to a decision for the above mentioned question.

Advantages

• Would provide an overview of how CS network is built - would be useful e.g. for marketing, seeing where local communities need to be jump-started etc

• The word of mouth within the (CS) Community is quite good. So quite often you hear about people before you met them the first time. There are also different groups and you could esily come behind a 'faked' rings of trust..

Disadvantages

• Privacy issues need to be considered carefully before starting off with this. Individual user's privacy settings need to be respected (and made clear to everyone before this would be launched). Who could/should use the data?
  • Anyone can go ahead and write a bot that parses people's profiles that are publicly visible.
  • On the other hand, how many people are able to do that? Compared to number of people who are able to click a link on your profile.

• Solutions
  • Different levels of access to this tool (anonymized overviews for everyone, in-depth views only for trusted people e.g. ambassadors who could use the data to do something meaningful in the real world?
  • Having option to turn it off for you, you wouldn't show up on networks of other people at all
  • Having only your friends above level X being able to check up your friendship network

Discussions

• trust system.. "How do I know..." 1
• trust system.. "How do I know..." 2

• effective tool for following purposes?... and more?

Comments

Add your comments here, sign using [[User:Guaka|Guaka]] 19:45, 14 August 2006 (EST)

• The "scientist" part of me digs this, and hopefully this is something I could start working with in Nelson (but only after bug-hunting, core feature development and whatever else needs to get done first :) Anu 10:38, 28 December 2006 (EST) - postponed until post-collective life is more settled and I get some core CS development issues more streamlined.
• I like that. --Alex 09:38, 11 January 2007 (EST)

See also