Do you work in a library which has either an online search or OPACs with a catalogue search, or similar?
I’ve started a Google Map with links to word clouds of users’ search keywords. The map so far (http://bit.ly/dE3hrh) has just one set of search keyword clouds – it would be great to have more from around New Zealand (and beyond).
What you need:
- Any kind of search tool or catalogue which produces a log of search keywords entered by users.
- To be able to nominate a geographic location for the dataset.
- Ideally – web statistics which include a list of the search keywords and (useful but not essential) their frequency. But as long as the data exists in some format (eg log files, or even just a list) it will still work.
If you’d like to contribute just email me (rebecca.cox @ Natlib) or comment here.
We collect web server log files and feed these into our web statistics software (Urchin, a version of Google Analytics which is installed and managed in-house.) From here you can export data in Excel format. I’ve cleaned this up, selected 500 terms from the top and bottom of the list, and created word clouds at www.wordle.net
Web stats give access to a wealth of data and can help identify audiences and behaviour which are not otherwise visible.
A while back, I checked the web stats for Papers Past to see how much “brand aware” search traffic the site was getting, and discovered there’s a significant number of people who appear to be searching the site for specific content using external search engines, eg site:paperspast.natlib.govt.nz “anti-opium association” or papers past deaths ashburton 1921.
You can look deeper by segmenting web stats by a range of criteria, from the number of words visitors use in their searches, to visitor domains (eg break out all the traffic from domains ending in .ac.nz), frequency of visit, number of pages viewed per visit, and more. For more on this, see Seb Chan’s Continuous Refinement and Data Driven Dynamic Personas from Webstock this year.
Another form of web visitor stats are heatmaps, which give a visualisation of where users are clicking on a web page (try Clickdensity or Clickheat). Here’s a heatmap showing the activity on our new homepage for the first few days after it went live.