A conference presentation at ASIST2010 about the (bad) idea of grouping web search thumbnails together at the top of Search Engine Results Pages. Just don't do it!
actually 3rd year undergraduate
to test a hypothesis
if honest - had chats about this *kind* of idea with friends at MSR once
flirting with the idea for ages
does it for you, if you really want it by default
improved both finding and re-finding
visual snippets can be smaller, and reveal content
work well for searching and re-finding
re-finding - woodruff et al 2001, teevan et al, 2009, etc
50% - tauscher & greenberg 1997, and 80% - cockburn and mckenzie 2001
- people re-find rather than book mark etc (teevan again)
pic faster than text (discussed by woodruff 2001) - also zheng’s chi 2009 paper (judgements)
Yahoo BOSS + ShrinkTheWeb API
should make optimal use of thumbnails
+ ideally shown during re-finding
+ grouped for quick comparison
+ all at top - so can judge quickly then search otherwise - dont have to scroll
+ Kaasten at al said 200px best for recognition - did 8 to page to fit
- but less results above fold
- also...
this didnt look the best - acknowledging limitations - bit unblanaced
lookup - population of brazil
subjective - choose jewelry
A-B were counterbalanced
well done to 3rd year
nothing really significant.
obviously a small study - maybe only a pilot study at best
only significant trend here is the diff in the subjective finding. where people seemed to be doing the judgement thing
but did get lots of insight into why
possibly type II error - because of low #Ps
the Revisit Rack was fast and efficient if the result was easily recognizable on that page of results
spatial disconnect
might want to study the impact of thumbnails on exploratory search
although could be more to do with the visual nature - couldn tell if judging website, or jewelry