Information Visualization: the Information Architecture Connection (Sherry Koshman, Ph.D.)
Session description
Raw notes to be polished:
Koshman, IA Program, U of Pitt
http://www.sis.pitt.edu/~skoshman
RESEARCH FOUNDATION
- interface design
- HCI
- Human Info Processing: how people use systems, effectiveness
- Information Retrieval
*Takes advantage of visual processing
Visualizations may help search w/
- how doc set relates to query
- re-prioritizing doc ranking
- improving cumbersome lists of docs
VISUAL PROCESSING CONCEPTS
- viz systems are intuitive
- uses low level preattentive visual processing
*Gestalt laws
- proximity: grouping based on spatial relationships
- closure: our ability to perceive boundaries
- continuity/connectedness: perceive things as more alike due to lines, _trumps proximity_
* it takes advantage of auto-processing and massively parallel processing in our visual cortex!
VIBE USER STUDY
- Experts/Novices
- Compare VIBE vs text interfaces
- Findings for group/task
- no significant difference between experts and novices
- visual recog faster than text processing?
*- need to think of new ways of testing these systems and viz field
SHARED CONCEPTS W/ IA
Common metaphor: space, structure
Provides high level and item level views
Visual org
Navigation
User-centered focus
Application to the Web
TOUCHGRAPH USER STUDY
- documented in Journal of Web Engineering
- similarity testing?
- like social network graphing, but applied to topics
- hub and spoke model
- change "radius" to set range of connections from a node
Conclusions
- Level of topic knowledge is a predictor on node use
- implications: knowledge diversity, does viz system handle gaps in knowledge of user
OTHER VIZ SYSTEMS
Mooter.com
*- clustering algo based on psych principles, not just mechanical
- same as Vivisimo? Clusty?
Kartoo.com
*- represent viz challenge of # of arbitrary meanings that need to be learned (e.g., icons)
- cartographic map metaphor
Grokker (trial software)
- Clustering
- "solar system", "planetary" approach
Stamen.com
- take nouns from GNews and map as _sparkline_ (Tufte!)
- different-colored chips; size = extent, color = trend (gain, lose coverage)
- rollover a chip: small histogram of coverage over the week
MapStan Search
- Document ranking
- streetmap metaphor
- a circle is a _town square_ (Ha.)
IMPLICATIONS FOR WEB SITE DESIGN
Interface design
- document glyphs (iconic representations) may be personalized by users (Wist!)
- Viz type: matrix of models vs features to build toolset (how about let user choose viz type too?)
- Use of text
Web IR
- viz of user activity (e.g. user query logs)
- viz content activity (e.g., what's been updated? etc., RSS feeds)
- viz of site content maps
- viz of controlled vocab (Visualthesauraus.com, thinkmap) and folksonomies (social network mapping?)
*- perhaps having a viz system do all the work of info retrieval is asking too much: need IA! (stained glass windows may have been great for communicating ideas during medieval times when few people were literate, but what about today?
Raw notes to be polished:
Koshman, IA Program, U of Pitt
http://www.sis.pitt.edu/~skoshman
RESEARCH FOUNDATION
- interface design
- HCI
- Human Info Processing: how people use systems, effectiveness
- Information Retrieval
*Takes advantage of visual processing
Visualizations may help search w/
- how doc set relates to query
- re-prioritizing doc ranking
- improving cumbersome lists of docs
VISUAL PROCESSING CONCEPTS
- viz systems are intuitive
- uses low level preattentive visual processing
*Gestalt laws
- proximity: grouping based on spatial relationships
- closure: our ability to perceive boundaries
- continuity/connectedness: perceive things as more alike due to lines, _trumps proximity_
* it takes advantage of auto-processing and massively parallel processing in our visual cortex!
VIBE USER STUDY
- Experts/Novices
- Compare VIBE vs text interfaces
- Findings for group/task
- no significant difference between experts and novices
- visual recog faster than text processing?
*- need to think of new ways of testing these systems and viz field
SHARED CONCEPTS W/ IA
Common metaphor: space, structure
Provides high level and item level views
Visual org
Navigation
User-centered focus
Application to the Web
TOUCHGRAPH USER STUDY
- documented in Journal of Web Engineering
- similarity testing?
- like social network graphing, but applied to topics
- hub and spoke model
- change "radius" to set range of connections from a node
Conclusions
- Level of topic knowledge is a predictor on node use
- implications: knowledge diversity, does viz system handle gaps in knowledge of user
OTHER VIZ SYSTEMS
Mooter.com
*- clustering algo based on psych principles, not just mechanical
- same as Vivisimo? Clusty?
Kartoo.com
*- represent viz challenge of # of arbitrary meanings that need to be learned (e.g., icons)
- cartographic map metaphor
Grokker (trial software)
- Clustering
- "solar system", "planetary" approach
Stamen.com
- take nouns from GNews and map as _sparkline_ (Tufte!)
- different-colored chips; size = extent, color = trend (gain, lose coverage)
- rollover a chip: small histogram of coverage over the week
MapStan Search
- Document ranking
- streetmap metaphor
- a circle is a _town square_ (Ha.)
IMPLICATIONS FOR WEB SITE DESIGN
Interface design
- document glyphs (iconic representations) may be personalized by users (Wist!)
- Viz type: matrix of models vs features to build toolset (how about let user choose viz type too?)
- Use of text
Web IR
- viz of user activity (e.g. user query logs)
- viz content activity (e.g., what's been updated? etc., RSS feeds)
- viz of site content maps
- viz of controlled vocab (Visualthesauraus.com, thinkmap) and folksonomies (social network mapping?)
*- perhaps having a viz system do all the work of info retrieval is asking too much: need IA! (stained glass windows may have been great for communicating ideas during medieval times when few people were literate, but what about today?
posted at 3/07/2005 08:30:00 AM
<< Home