The methodology being used is described in this post.
You are invited to comment and criticise as much as you like :)
Semantic web visualisation for Gordon Brown
Semantic representation for Nick Clegg
Semantic representation for David Cameron
Wednesday, April 21, 2010
Day 2 Semantic analysis of the UK general election
This is the first daily semantic wall about three political leaders David Cameron, Conservative; Gordon Brown, Labour and Nick Clegg, LibDem.
The methodology being used is described in this post.
You are invited to comment and criticise as much as you like :)
Semantic web visualisation for Gordon Brown
The methodology being used is described in this post.
You are invited to comment and criticise as much as you like :)
Semantic web visualisation for Gordon Brown
Subscribe to:
Post Comments (Atom)
I am not very well versed in the discipline of semantics, but to me it's all about meaning. What you are getting out of this analysis is a just big list of words.
ReplyDeleteThose tag clouds do not seem to be telling us anything about the meaning of what's being said online, they're just telling us what words are being used in association with the party leaders.
We have to think for ourselves about the meaning of those words that are being commonly used. And your automatic process has stripped away the contextual information that would enable us to do it.
Derek, what I can see and what you can't see is the ability to drill down to the individual citations relevant to each concepts in the wall.
ReplyDeleteThe ability to compare and contrast the semantic 'word map' has some advantages and, of course, we are able to use standard databases to identify the flow of concepts (and daily/competitor unique's). This methodology also has the advantage of showing changing trends and therefore has the potential to forecast with some degrees of accuracy. See Paul Dagum and Adam Galper "Time-series prediction using belief network models" at the Semantic Portal Wiki http://tw.rpi.edu. They proposed a Bayesian solution to make the information forecast more accurate. As it happens I do have a Bayesian logic software engine but have not yet applied it to this research project.
I used the semantic wall form of visualisation because I had it available and it is not too much data for a casual observer of a complex experiment to observe.
As you will see, we do not use word clouds which is a count of all words in a corpus based on the research that Bruno Amaral published at the Bledcom conference last July. Our methodology is based on semantic concepts identifies in each citation.
So we have not stripped away the contexts There are many) and, of course, can recover a wide range contextual data.