Transcript chp15
Chapter 15: Information
Search & Visualization
Team 3: Jacob Hicks, Victor Chen, Saba Alavi
Introduction
Information exploration overload/anxiety?
Object-actions Interface (OAI) model helps by:
N00bs in an information-exploration system…
separating different task concepts
separating high-level interface issues from low level interface
issues
struggle to understand what they see whilst remembering their information
needs
might be distracted by learning complex query languages/elaborate shapecoding rules
need direct-manipulation designs/simple visual-coding rules (low cognition)
can request additional features by adjusting control panels
Experienced users want more functionality and power:
a wider range of search tools, lots of options
Introduction (cont’d)
Task objects represented by interface objects in
structured relational databases or text/media document
libraries
Structured relational databases made up of relations
and a schema (model) to describe relations
Relations have items (tuples/records), which consist of
multiple atomic attributes, each of which have attribute
values
Textual document library comprised of collections and
descriptive attributes (e.g. location, media type,
curator, donor, etc.)
Introduction (cont’d)
Multimedia document library same as
textual document library, only instead of text,
it’s media: images, sound, video, animations,
etc.
Task actions (i.e. fact finding) decomposed
into browsing/searching, represented by
interface actions (i.e. scrolling, zooming,
joining, linking)
Finding aids help users focus their info needs
(i.e. table of contents, indices, abstracts, etc)
Database Query/Phrase Search
SQL a widespread standard for searching in
structured relational database systems
Requires substantial time investment to learn
Computer’s capacity for responding to natural
language query often limited
Tradeoff exists between ease of use and
usefulness
Empirical studies illustrate better performance
and more satisfaction when users are able to
view and control the search
Database Query/Phrase Search
Improved designs & consistency across
differing systems allows for faster
performance, fewer mistakes, and more
successful searches
Recommends four phase framework:
Formulation – expressing the search
Initiation of the action – launching the search
Review of results
Refinement – formulating the next step
Multimedia Document Searches
Current approaches to locating media rely on parallel databases
and document searches
Advocates for ambitions captioning and attribute recording
Classification according to useful search categories useful,
though costly and imperfect
Graphical specification of query components:
Photo search
Map Search
Design/diagram search
Sound search
Video Search
Animation Search
Information Visualization
Bandwidth
of vision is high
Overview first, zoom and filter, then
details on demand.
Data type by task
1-D Linear
Text
documents
Source Code
Bifocal Display
Value Bars
2-D Map
Maps
Floorplans
Newspaper
layouts
3D World
Real
objects, models, ect.
Must keep track of position orientation
Occlusion
Temporal data
Time
lines
1D linear
Start and finish time
Events may overlap
Multidimesional data
n-dimensional
space
Databases with n attributes
Can be 2D or 3D
Scattergrams
Tree data
Hierarchies
Can
be shown as lines and nodes
Tabbed text files
Cones in 3D
Network data
Networks
Cannot
be written as a tree
Node-and-link
Square matrix
Overview Task
Movable
field of view
3 to 30 zoom amount
Fisheye
Zoom task
View
a specific area in detail
Smooth zooming preserves orientation
“A satisfying way to zoom in is to point to
a location and to issue a zooming
command”
Filter task
Remove
unwanted items
Widgets to regulate process
Dynamic control of items
Details-on-demand task
Select
item or group to get details
Click on an item to get popup window
Relate task
View
relationships amoung items
Select an item to highlight related items
History task
Keep
history to support undo
Tasks from the past combinded
Extract Task
Extraction
of subcollection of parameters
Allow to save the records that result from
a search
Save settings
Advanced Filtering
Dynamic queries
Numeric range sliders
Alphasliders for names
Bottons for small sets of categories
Commercial Information–retrieval
systems
Example – DIALOG or First Search
Permit complex Boolean expressions
with parentheses but they are difficult to
use
When we say or in English it means
not both, but in Boolean OR is inclusive .
New York and Boston ( result 0 )
Another form of filtering …
Apply a user-constructed set of
keywords to dynamically generated
information. Such as incoming email
messages…..
A social form of filtering is collaborative
filtering ….. Music, Restaurants ..
Summary
Improved user interface to traditional
databese-query or multimedia-document
search will spawn appealing new products.
The more Flexible the better…
15.7 Search in complex structured
documents. graphics, images, sound or
video persents grand opportunities for the
design of advanced user interfaces and
powerful search engines .