Please enable JavaScript.
Coggle requires JavaScript to display documents.
8 [IR] Interactive IR (Query Formulation and reformulation (More recently,…
8 [IR] Interactive IR
Query Formulation and reformulation
user's goals: Identify the right query for current need
How
infer the right query terms
infer the right composition of terms
More recently
Spoken queries
Queried from scanning/taking pictures
search by drawing
What IR system can do?
Ask more from the user
encourage long/complex queries
obtain contextual information from users
Infer from relevant items
infer from previous queries, term auto suggestions
query reformulation: historical queries/related queries/ providing facets
infer from previous retrieved documents
Design interactive IR
pay attention to usability
Learnability
Efficiency
Memorability
Errors
Satisfaction
General design principles (design desiderata for search user interfaces)
offer informative feedback
Support user control
Reduce short-term memory load
Provide shortcuts for skilled users
Reduce errors; offer simple error handling
Strive for consistency
Permit easy reversal of actions
Design for closure: user has a sense when the search is complete
General Design Principle(Offer efficient and informative feedback)
show search results immediately
show informative document surrogates; highlight query terms
Allow sorting of results by various cretiria
show query terms suggestions
provide simple and meaningful history mechanism
Integrated search and navigation
Evaluating interactive IR systems
three dimensions of usability
effectiveness
: accuracy and completeness with which users achieve specified goals
efficiency
: resources expended in relation to the accuracy and completeness with which users achieve goals
Satisfactory
: freedom from discomfort, and positive attitudes towards the user of the product
evaluation methods
standard IR metrics -- precision, recall, nDCG
Informal usability test -- interview, filed study in early stage
Formal usability test -- controlled user experiment
search log analysis and A/B testing
Controlled user study
techniques in user study
IRB
Sonsent form
training task
pilot study
balance condition ordering effects
removing learning effects, fatigue effects and task effects
latin square ordering
obtaining particepants
obtain participants subjective responses
congnitive load
satisfactory
Search result presentation
Post-search analysis
depends on the types of information needs
simple cases: identification of the relevant items
complicated cases: exploration, learning and investigation
Post-search analysis types: looking for information to explore
trends
comparisons
aggregation and sacling
identifying a crtical subset
assessing
interpreting
others: cross-reference, summarize, find evocative visualizations
Document surrogate
design of document surrogate influences user clicks
KWIC: key word in content