[Dspace-general] Wordle visualization of DSpace content

[Robin Taylor]

Hi Bram,

More fluff for the 'fun on Friday' category - I was asked to generate a dynamic Wordcloud of search terms entered into our IR to be flashed up on a big screen in our library. If you interested you can see it at http://www.era.lib.ed.ac.uk/searchQuery (** please use Mozilla as that's what its designed for). As a piece of 'art' its rubbish in comparison with what Wordle can produce, the only interesting thing to come out of the exercise for me was the discovery that 99% of our searches come from federated search engines rather than being entered directly via the UI.

Cheers, Robin.
  

Robin Taylor
Main Library
University of Edinburgh
Tel. 0131 6513808  

> -----Original Message-----
> From: dspace-general-bounces at mit.edu 
> [mailto:dspace-general-bounces at mit.edu] On Behalf Of Bram Luyten
> Sent: 17 July 2009 14:00
> To: dspace-general at mit.edu
> Subject: [Dspace-general] Wordle visualization of DSpace content
> 
> Hello,
> 
> In the category, fun on friday, I was curious to investigate 
> the results of feeding DSpace item titles into Wordle ( 
> http://www.wordle.net ), and see what would come up.
> 
> Wordle visualizes the occurrence of words for any amount of 
> text you feed it. Basically Worlde counts the times a 
> specific word occurs, and represents words that occur many 
> times large, and words that only occur a few times, smaller, 
> in one resulting picture.
> 
> As a data source, I used K.U. Leuven's LIRIAS repository ( 
> http://lirias.kuleuven.be ), a large and rapidly growing 
> repository. This DSpace's hierarchy is subject oriented, as 
> the communities and collections are organized according to 
> the institution's organizational structure. For this 
> experiment, I took three top level communities: the 
> Biomedical Sciences group, the Humanities and Social Sciences 
> group and last (but not least) the Sciences, Engineering and 
> Technology group.
> 
> Using @mire's reporting suite ( 
> http://atmire.com/USB/resources/reporting_suite.html ) it 
> took me five minutes to generate a clean list of the item 
> titles of International Publications (a small subset of the 
> content) for each of these top level communities, that were 
> submitted in 2009 (500+ for each of these groups).
> 
> These lists were used to create following Wordles:
> Humanities and Social Sciences - 
> http://www.wordle.net/gallery/wrdl/1003572/K.U._Leuven_Humanit
> ies_and_Social_Sciences_publications_2009
> Biomedical Sciences - 
> http://www.wordle.net/gallery/wrdl/1003562/K.U._Leuven_Biomed_
> Publications_2009
> Science, Engineering and Technology - 
> http://www.wordle.net/gallery/wrdl/1003577/K.U._Leuven_Science
> %2C_Engineering_and_Technology_publications_2009
> 
> It was funny to see that almost all titles were in english 
> for the Biomed and SE&T groups. For Humanities and Social 
> Sciences, there was a mix between english and dutch titles. 
> Wordle allows you to filter the most common words (the, an, 
> a, ...) for one particular language. So to clean the 
> Humanities & Social Sciences Worldle from both english and 
> dutch stop-words, I had to do some manual work on the list. 
> 
> Although already a sub-selection of three groups was made, 
> you still see a lot of "generic" scientific terms, and not so 
> many interesting subject keywords. That's quite logic, 
> because although the scientists belong to the same group, 
> they're still dealing with a variety of subjects.
> 
> When zooming in on more specific subjects, here's the Wordle 
> from the Computer Science department 2009 publications (one 
> subcommunity level below the Groups):
> http://www.wordle.net/gallery/wrdl/1003647/K.U._Leuven_Compute
> r_Science_publications_2009
> 
> And even more specific, here's the one for the researchgroup 
> of Experimental Radiotherapy, under the Department of 
> Oncology in the group of Biomedical sciences. For this one, I 
> took all of the publications from 2000-2009 to get a relevant 
> selection.
> http://www.wordle.net/gallery/wrdl/1003638/K.U._Leuven_Experim
> ental_Radiotherapy_Publications_2000-2009
> 
> best regards,
> 
> Bram Luyten
> 
> @mire - http://www.atmire.com
> 
> Technologielaan 9 - 3001 Heverlee - Belgium
> 533 2nd Street - Encinitas, CA 92024 - USA
> 
> http://www.togather.eu - Before getting together, get Tog at ther 
> 
>


-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.