- BCS IRSG

advertisement
The context of the interface
Ian Ruthven
University of Strathclyde
interface and interaction
• interaction
– designing for reciprocal behaviour of person and system
– largely influenced by cognitive models of behaviour
– largely (I think) we often pay attention to components
of interaction than whole interaction
• interface
– means of interaction – how interactive possibilities
manifest themselves
– means of understanding how to behave
– largely, now, means of popularity and uptake
• want to look at the interface in more detail and the
possibilities that interface design offer us
2
interface and interaction
• interaction
– designing for reciprocal behaviour of person and system
– largely influenced by cognitive models of behaviour
– largely (I think) we often pay attention to components
of interaction than whole interaction
• interface
– means of interaction – how interactive possibilities
manifest themselves
– means of understanding how to behave
– largely, now, means of popularity and uptake
• want to look at the interface in more detail and the
possibilities that interface design offer us
3
models in information seeking
• information seeking
can be
– cognitively demanding
– frustrating
– just plain difficult
• but can also be
– stimulating
– fun
– engaging
4
5
6
7
‘regular’ IR
8
‘regular’ IR
9
‘regular’ searching
• in interactive terms
– no identification of individual
– high prediction of output and interaction (generic and
obvious interaction models)
– relatively short term interaction
• not always but drive to easy satisfaction
– values direct interaction
• emphasis on query
• and verbal (textual communication)
• not always suitable for complex interaction
– tendency to ‘linear logic’ in interaction
• e.g. documents retrieved contain query terms
• structured output
10
‘regular’ searching
• ranking model very dominant
– useful model for users (at least to increase
efficiency) and for some search tasks
– IR has constructed a lot around ranking
approach
•especially in evaluation
– but perhaps too much
•personalisation in particular
•even work on implicit feedback tended to
concentrate on improving rankings
11
contextual searching
12
contextual searching
• in interactive terms
– identification of individual
• but also group (collaborative filtering)
• some personalised interaction
– takes time to create relationships
• not just recommendations but reviewers
– relatively longer term interaction
• and interface support for increased engagement
– values implicit interaction
• emphasis on browsing
• and non-verbal communication
– tendency to ‘spiral logic’ in interaction
13
interface
• affordances
– possibilities made available (or at least perceived
abilities)
• also used for interfaces that suggest/encourage particular
interaction
• still (I think) under-investigated in IR
• in IR (I think)
– interactive work has laudable aim of making interaction
easier cognitively
• (relevance feedback, interactive query expansion, metaphors
such as document piles)
– but not really tackled making search more enjoyable or
interesting
• motivated searchers always do better
– not suggesting we YouTube everything
14
persuasive interaction
• interfaces that encourage (persuade) people to
behave differently
– either macro-persuasion (stop smoking, drive more
carefully,..) or micro-persuasion (customer reviews,
‘more like this’)
• many techniques, including
– reduction – making complex tasks simpler (Google
suggest vs interactive query expansion)
– tunnelling - (leading users into a step by step process or
particular strategy, e.g. dialogue models of search)
• although may not be an effective strategy!
• force vs persuasion
– self-monitoring – giving feedback to users (e.g. Query
Performance Analyser)
15
persuasive interaction
• can we design interfaces that help people
interact more usefully?
– make interaction (that we believe to be useful)
more enjoyable and likely to happen?
– either
•in a ‘big’ sense – whole interaction model
•or in ‘small’ ways – individual interactive tasks
16
17
18
19
emotional
• emotional aspects of searching still underinvestigated
– not only what we do with emotions
• we can recognise emotions
– but how to design to help create emotions
• different emotions have different effects
– can help people be more creative
– can help people focus
• different emotions can be imaginatively used in different search
situations
20
emotional design
• Don Norman
– visceral level of design (appearance)
– behavioural level of design (pleasure and effectiveness
of use)
– reflective level of design (self-image, personal
satisfaction, memories)
• argues, in part, that objects designed to be emotionally
appealing are more likely to be used
• but also more likely to forgive an object that does not work well
if we like it!
• this is important for an area like IR
– especially for more difficult tasks
– and tasks that are unlikely to work every time
21
engagement
• functionality is important
– in IR we have tended to concentrate on functionality
and effectiveness
– less on usability (although interlinked with
effectiveness)
– still less on engagement
• engagement, of course, not a substitute for effectiveness
• but tendency to see IR as an intellectual challenge
– O’Brien and Toms
• useful language to talk about engagement
• need to respond to this with new designs
22
engagement
•
•
•
•
very important commercially
there is a flip-side for academic researchers
users do bring expectations
design and perceived quality is increasingly
important
23
summary
• interactive IR has created some great
systems
• we still have many challenges
– creating complex search requests
– analysing complex search results
• lots of very useful ‘under-the-bonnet’ work
– perhaps still not quite sure what to do with it!
• but need to consider excitement and
pleasure of search
24
Download