decorative tag cloud

It’s not often that I’d consider adding pure “eye candy” to the OPAC, but I couldn’t decide what would be the best way of making this tag cloud functional. So, I made an executive decision and decided it shouldn’t be functional 😀
If you run a keyword search on our OPAC, at the foot of the page you should see a keyword cloud (it might take a few seconds to appear). The cloud is generated from previous keyword searches used on our OPAC. Here’s the one for “library“…
For multi-keyword searches, an electronic coin is tossed and you either get a cloud of the union or the intersection of your keywords. The former uses previous searches that contain any of the keywords, and the later is only those that contain all of them (if that makes sense!)
As it’s not functional, the cloud is just a decorative window into the hive mind of our users.
I’m interested to hear what you think — should the cloud be functional, or does it work as just “eye candy”?

12 thoughts on “decorative tag cloud”

  1. Thinking about it more, I might tweak it so that clicking on a keyword in the cloud gives you the option of either combining that with your current search terms (if that would actually give results), or starting a new search with just that keyword.

  2. I’m not quite following, Dave. The terms in the tag cloud are the result of previous multi-word searches that contained one or more of the user’s terms?

  3. That’s right — it’s looking at a database of around 2 years worth of previous keyword searches.
    For multi keyword searches, you either get a cloud generated from one or more of your search terms (union/boolean OR) or from all of them (intersection/boolean AND). With the latter, the cloud shrinks as you add more terms to the search. With the former, it grows.
    I’ve decided to make the cloud keywords clickable just to see if anyone will use them.

  4. For the (very) (too) high percentage of students who have no idea how the catalogue works it would be great … put in word and magical suggestions that are kind of relevant, and useful, display.
    It’s as good as anything else that actually makes sense to us but is mystical to the students. (eg. the “relevance ranking” on the III catalogue at my uni…. scrap that, it doesn’t make sense to me either)
    It would drive me (and other librarians) bonkers. I’d keep trying new searches to try to work out what was triggering the cloud… classic variable ratio reinforcement – shown in psychological trials to reinforce obsessive OPAC search behaviour in 9/10 librarians.
    Good thing that we are trying to serve the students not the librarians. I like the idea. A lot. Decoration and functionality via lateral thinking and electronic coin tossing.
    Now … can you work on randomly assigning colours? Maybe according to the first letter in the word (so A-C becomes red, D-F becomes orange etc)… and see how long it takes ppl to work out. 🙂

  5. Rainbow tag cloud? I like 🙂
    The keyword cloud is taking a little too much CPU time to generate at the moment, so I need to do a little bit of creative recoding.
    I’ve also been playing around with the WWW::Wikipedia Perl module. It has a handy “related” feature that gives you all of the hyperlinked terms in a specific Wikipedia article (these could be used to seed a list of related terms). To get the most relevant Wikipedia article for the OPAC search terms, the easiest thing to do is to use the Yahoo-Search module and limit the search to just the Wikipedia site (“”). The first Yahoo result returned should be a link to the most relevant article.
    As an example, a search for “Fremantle” brings up a list of 91 related terms including: AC/DC, Bon Scott, Croatian people, Fire-stick farming, Gage Roads, James Stirling, microbrewery, Nyungar, psychiatric hospital, Subiaco Oval, The Amazing Race, Western Australian Maritime Museum, White Gum Valley, and Xanthorrhoea.
    The $64 question is how would some of our academic staff react if the library began integrating Wikipedia sourced content into the OPAC? Perhaps it’s a good job I don’t work at the Univeristy of Brighton, or I’d have Tara Brabazon calling for my scalp 😉
    Anyway, I’ve got clouditis at the moment, so I added a cloud of the most popular recent search terms to the front page of the OPAC yesterday (it refreshes the list every 30 seconds).

  6. Huh! You just listed some of the best things about where I live – my suburb is next door to White Gum Valley and I’m in walking distance of Bon Scott’s grave…. and then you spoiled it by mentioning another product of this area – Tara B. is a graduate from the university where I work.
    Couple of weeks ago a mob of librarians at work discovered the article from January ( White Bread For Young Minds ). They were very supportive of the ideas – especially that the undergrads are only allowed to quote from the list of articles that she provides them.

  7. In the name of “perpetual beta”, I’ve temporarily replaced the tag cloud with a list of terms from Wikipedia just to see how well (or not) that works. As hinted in a previous comment, it uses Yahoo to try and locate the most relevant Wikipedia article for the OPAC search terms used.
    Rather than show all of them at once, it picks a random list of 8 to display (with the option of viewing more).

  8. I love the idea of tag clouds working within the OPAC. Being a visual person myself, I relate well to them and hope to deploy them at my library some day. I see this kind of technology as just one more way to assist users, especially the ones that are reaching us from off campus.
    I saw a demonstration of AquaBrowser Library at CiL 2008. I was fascinated, but saw some issues right away. Mainly, it needed to be explained in some detail before most of us understood how the tag cloud was being generated and the relationship of the cloud results to the search that initiated it. Have you had any users asking what the heck is this thing? What kind of use of the tag cloud are you seeing?
    Great job!

  9. Thanks Tim!
    The cloud on the front page of the OPAC is generated from analysis of all of the successful keyword searches from the current and previous day. It will vary from hour to hour, but at the moment, a keyword has to been used around 100 times or more to appear.
    No feedback yet from users, although I’m monitoring usage and I’ll post some stats soon.
    A nice unintended benefit of adding the Wikipedia content is that it’s generating suggestions for questions – e.g.:
    what is the answer to the ultimate question of life, the universe and everything?
    what is the capital city of France?
    why is the sky blue?

  10. Just had a quick look at the usage stats. Since Monday, we’ve had 86 clicks on Wikipedia terms and 82 clicks on keywords that appear in the cloud on the front page.
    Not particular high figures, but we’re in a fairly quiet period in the library at the moment. As a benchmark, we’ve had 901 clicks on “did you mean” spelling suggestions and 295 on “people who borrowed this, also borrowed” suggestions this week.

Comments are closed.