30 September 2010

Talking about data visualisation



Good podcast by Involution Studios, including a discussion about data visualisation, which works remarkably well despite being in audio only, because of the index of links Involution provide. The three discussants (Brian Staats, Irene Ros and Michael Dila) covered the opportunities for giving access to public information (and also obfuscating it) that data visualisation brings, the varied backgrounds (often more statistical than design-based) that data visualisers have, the need for people to be educated to be able to read visualisations critically. Their discussion, of course, mentioned both Edward Tufte and Hans Rosling and led me to Rosling's recent, wonderful visualisation of population growth, using IKEA storage boxes (above).

From the podcast I linked to a talk by Nick Felton (of the Feltron Annual Report), information designer who has collaborated Rob Deeming and Ken Riesman (with backgrounds in natural language processing and algorithmic methods) on a project called 'What we are saying', comparing (and visualising) the front page stories of the NY Times and other news sources with user-generated content on Twitter and blogs. Some of their less expected findings for me were that whereas coverage of front page stories is pretty much evenly shared across topics, user-generated content has much stronger biases (in the sample week they covered, healthcare was the topic for 27% of UGC, compared to 19% of news stories) and that, generally, UGC expresses more positive sentiment than the news (maybe this is less suprising).

(Note Ken Riesman has created Pluribo, which looks at Amazon reviews of products and produces a short, digestible precis. Unfortunately I couldn't get it to work. Would be a wonderful tool for managing reading certain kinds of UGC if it did.)

No comments:

Post a Comment