I've been using Gephi to model topics as produced by topic modeling software in text analysis and I've been blown away visually by the results. If there are any text analysis or computational linguists who have done similar work, I'd love to hear about it.
https://dhs.stanford.edu/visualization/topic-networks/
Topic Networks
-
- Gephi Community Manager
- Posts:964
- Joined:09 Dec 2009 14:41 [phpBB Debug] PHP Warning: in file [ROOT]/vendor/twig/twig/lib/Twig/Extension/Core.php on line 1275: count(): Parameter must be an array or an object that implements Countable
Re: Topic Networks
wow, it rocks!
Re: Topic Networks
It just so happens to turn out that the MALLET topic analysis package outputs data that is remarkably suitable for network representation. I'm writing up the whole thing with a host of images and I'll be releasing the original MALLET data as well as the various .gephi projects files from the two drafts and the final so that other people can play around with a very interesting dataset. I should be putting it up on Friday or Monday, depending on a few other things.
I basically started it just to better understand how topic modeling works, but there are some rather visually striking results.
I basically started it just to better understand how topic modeling works, but there are some rather visually striking results.
-
- Gephi Community Manager
- Posts:964
- Joined:09 Dec 2009 14:41 [phpBB Debug] PHP Warning: in file [ROOT]/vendor/twig/twig/lib/Twig/Extension/Core.php on line 1275: count(): Parameter must be an array or an object that implements Countable
Re: Topic Networks
Cool, release also the graphs in .gexf as it is a format more stable than the .gephi.
-
- Gephi Community Manager
- Posts:964
- Joined:09 Dec 2009 14:41 [phpBB Debug] PHP Warning: in file [ROOT]/vendor/twig/twig/lib/Twig/Extension/Core.php on line 1275: count(): Parameter must be an array or an object that implements Countable
Re: Topic Networks
You could even submit the images at the Art & Science Exhibition Proposals - ISMB/ECCB 2011
http://www.iscb.org/ismbeccb2011-submis ... exhibition
http://www.iscb.org/ismbeccb2011-submis ... exhibition
Re: Topic Networks
I'd love to but it looks like that's only open to conference participants...
Over the weekend I've been playing even more--you guys have really outdone yourself with providing the ability for scholars to visualize their data in the most captivating ways.
Over the weekend I've been playing even more--you guys have really outdone yourself with providing the ability for scholars to visualize their data in the most captivating ways.
-
- Gephi Community Manager
- Posts:964
- Joined:09 Dec 2009 14:41 [phpBB Debug] PHP Warning: in file [ROOT]/vendor/twig/twig/lib/Twig/Extension/Core.php on line 1275: count(): Parameter must be an array or an object that implements Countable
Re: Topic Networks
I'm sure they will listen to youIf you can not attend the conference but still have artworks to present on the Art & Science exhibition please contact artandscience@ismbconf.org for further information.
Re: Topic Networks
For those interested, I've completed my little project and the results are here:
https://dhs.stanford.edu/comprehending- ... umanities/
Of particular note is the data page, which has links to a pair of gexf files for two iterations of the topic network (unfortunately the middle draft was corrupted).
https://dhs.stanford.edu/comprehending- ... ties/data/
I would love feedback as to the process and my description of it. While I feel very comfortable with how I've described things, if I've made a mistake or exaggerated something or simply misspelled something, please do let me know. As usual, thanks for designing the tool.
Please keep in mind that this was not a long-term research project but rather a short term exploration and demonstration, so while I would very much appreciate any methodological or theoretical criticism, I hope you'll note that I was very clear about the limited nature of my claims.
https://dhs.stanford.edu/comprehending- ... umanities/
Of particular note is the data page, which has links to a pair of gexf files for two iterations of the topic network (unfortunately the middle draft was corrupted).
https://dhs.stanford.edu/comprehending- ... ties/data/
I would love feedback as to the process and my description of it. While I feel very comfortable with how I've described things, if I've made a mistake or exaggerated something or simply misspelled something, please do let me know. As usual, thanks for designing the tool.
Please keep in mind that this was not a long-term research project but rather a short term exploration and demonstration, so while I would very much appreciate any methodological or theoretical criticism, I hope you'll note that I was very clear about the limited nature of my claims.