A Typology of Search Behaviours
There’s a naive view of search which is that somebody is looking for a specific piece of information, they frame a query and then they are satisfied or unsatisfied by the search result. It’s a very widespread mental model, and it is premised on the notion that a searcher has a precise idea of what they want (search is fully intentional and there is a clear “target” for the query) and that they know when they are satisfied. This is of course a notion that suits the taxonomist, because the intentional/target approach says all you have to do is predict likely targets and the likely language of search and map this to your taxonomy language and concepts (the document managers can worry about how the content gets tagged to your concepts).
The reality is much messier of course – search is not a discrete, defined activity, its intent can be vague or well specified or anywhere in between. It can be done by well-informed people in the domain or by novices. Known initial goals can be distracted and replaced by serendipitous discoveries. What “satisfaction” means and how you’d measure it is anybody’s guess – it can vary according to context and random chance.
The naive view was challenged several years ago by Andrei Broder in his elegant paper on “a taxonomy of search” where he identified three main types of search purpose on the internet (navigational, informational and transactional) and actually put some numbers behind them.
This work has now seen some significant development in a recent blog post by John Ferrara, who has used a personas approach to develop a typology of search behaviours. He hasn’t got any data behind this yet, but it’s a big step in the direction of a search model that might be sophisticated enough to map against real people doing real tasks.
And our next task will be to look at how the typology works inside the firewall for enterprise search. No, it’s not the same as web search, and anyone out there who tells you it is essentially the same thing, is lying or ignorant. The information landscape is radically different inside the enterprise (and it’s not just because there’s less porn) the needs are more functional and transactional, the expectations are radically different, and the search drivers are more extrinsic than intrinsic.
4 Comments so far
- Patrick Lambe
Thanks John, very helpful of you!
- Doug Cornelius
Patrick -
I find the biggest difference of searching inside the enterprise (as compared to the internet) is the variety of sources and searches available. Information is usually balkanized into different systems, with different ways to search them.
This is my take on different types of searches for documents in a document management system:
http://kmspace.blogspot.com/2007/04/four-types-of-document-searches.html
I have found the four types of search (fetch, recall, research and precedent) to hold up in searching other systems.
- Patrick Lambe
Thanks for this Doug, I like the four-part typology and the way you link these to the different ways that enterprise search and DMS search function.
Page 1 of 1 pages
Comment Guidelines: Basic XHTML is allowed (<strong>, <em>, <a>) Line breaks and paragraphs are automatically generated. URLs are automatically converted into links.
This paper describes search behaviour in teleporting (one big jump) and orienteering (small steps):
http://nform.ca/blog/2007/05/orienteering-vs-teleporting
Information scent comes from the design perspective:
http://www.steptwo.com.au/papers/kmc_informationscent/index.html
Found another old paper on information foraging:
http://www.firstmonday.org/issues/issue5_2/choo/
Posted on February 04, 2008 at 06:05 AM | Comment permalink