Intelligent User Interfaces Intelligent User Interfaces Research GroupResearch Group
Directed by: Frank ShipmanDirected by: Frank Shipman
Visual Knowledge Builder (VKB): Visual Knowledge Builder (VKB): Supporting Personal CollectionsSupporting Personal Collections
Problem: information tasks require a combination Problem: information tasks require a combination of location, comprehension, and modificationof location, comprehension, and modification• Current systems have emphasized the location of Current systems have emphasized the location of
information (e.g. search engines, portals)information (e.g. search engines, portals)• There are few tools supporting the activity of There are few tools supporting the activity of
comprehending and modifying the found contentcomprehending and modifying the found content
Approach: Spatial hypertextApproach: Spatial hypertext• Spatial hypertext expresses inter-document Spatial hypertext expresses inter-document
relationships via visual and spatial cuesrelationships via visual and spatial cues• Users develop personal visual languages during the Users develop personal visual languages during the
course of their activitiescourse of their activities• The system can recognize parts of the visual language The system can recognize parts of the visual language
and interpretation to support the user’s task. and interpretation to support the user’s task.
Frank ShipmanFrank Shipman
VKB
Expressing Relations VisuallyExpressing Relations Visually
VKB provides: A hierarchy of two-dimensional workspaces called
collections for placing information Easy manipulation of
visual properties of information
Information objects pointing to external content
Attribute/value pairs for attaching metadata
VKB
Experience with VKB UseExperience with VKB Use
VKB
Project Management
ConferencePlanning
Writing Papers & Creating Presentations
VITE : Manipulating Structured VITE : Manipulating Structured Information in a Visual WorkspaceInformation in a Visual Workspace
Problem: computers generate/require lots of Problem: computers generate/require lots of structured information but people prefer natural structured information but people prefer natural representationsrepresentations• Incomplete representation –Incomplete representation –
formal representations abstract real contentformal representations abstract real content• Insufficient intermediate representation –Insufficient intermediate representation –
formal representations miss transient states during formal representations miss transient states during problem solvingproblem solving
Approach: 2-way mappings in a visual workspaceApproach: 2-way mappings in a visual workspace• Two-ways: (1) visualization and (2) visual parsingTwo-ways: (1) visualization and (2) visual parsing• Editable visualization and mappingEditable visualization and mapping• Direct manipulation of content in the workspaceDirect manipulation of content in the workspace
Haowei HsiehHaowei Hsieh
VITE
Problem Solving in a Visual WorkspaceProblem Solving in a Visual Workspace
Visual mapping is created in the Mapping Designer. Structured information is rendered as information objects in the workspace.
Users accomplish tasks by adjusting mapping designs and then manipulating information objects in the workspace.
VITE
Generating Mappings with EaseGenerating Mappings with Ease
The Mapping Assistant generates mappings based on:• a brief description of the
user’s task,• a statistical analysis of the
data set, and • a mapping design
knowledge-base combining results from the VITE evaluation with general graphic design principles.
The Mapping Assistant helps users generate initial mappings quickly so users can start working on the task right away.
VITE
GRC: Graphical Requirements GRC: Graphical Requirements CollectorCollector
Problem: Software requirements elicitation• Questionnaires enable feedback from a large
group but do not elicit rich design information.• Interviews and observations generate rich
content but are expensive in time and money.
Approach: Collect design information from probable end-users through:• End-user graphical user interface mock-ups• Textual descriptions and rationales for the
widgets and windows that they create
J. Michael MooreJ. Michael Moore
GRC
Examples of End-user Mock-upsExamples of End-user Mock-ups
Above: Other users relied more on Above: Other users relied more on textual description. The text dialog textual description. The text dialog
explains the role of the “Design” button.explains the role of the “Design” button.
Below: Some users depend Below: Some users depend heavily on graphical organization.heavily on graphical organization.
GRC
Generating Design Information from Mock-upsGenerating Design Information from Mock-ups
Term-frequency view of mock-up designs
Main analysis interface provides access Main analysis interface provides access through:through:• Filtering mock-up componentsFiltering mock-up components• Grouping components based on visual Grouping components based on visual
and textual similarityand textual similarity
Algorithms analyze spatial, textual, Algorithms analyze spatial, textual, and temporal information to and temporal information to generate pattern-based views of generate pattern-based views of design data to help construct design data to help construct domain modelsdomain models
GRC
MASH / WARPMASH / WARPLuis Francisco-RevillaLuis Francisco-Revilla
MASH
Problem: information needs to be adapted based Problem: information needs to be adapted based on the use contexton the use context• Information systems can employ multiple models in Information systems can employ multiple models in
order to adapt content and presentation.order to adapt content and presentation.• Conflict may occur as different models propose Conflict may occur as different models propose
contradicting suggestions.contradicting suggestions.
Approach: identifying mechanisms for dealing Approach: identifying mechanisms for dealing with conflictswith conflicts• Deliver a flexible context-sensitive solution to this issue
within the field of adaptive spatial hypermedia• MASH (Multi-model Adaptive Spatial Hypermedia) is a
framework for dynamic and adaptive behaviors.
Architectural Framework
Spatial Hypermedia
Platform
SpatialParser
Spatial Analyzer
Spatial Transformer
Composites
Metrics
Transformations
Behaviors
M1
M2
Mn
Spatial Hypermedia Generator
Contents
Dynamic
Adaptive
Generative
Interactive
Models
The MASH framework consists of three parts: a high-level abstraction of objects and relationships, a generic architectural framework, and a theoretical ontology of spatial adaptations.
Ontology of Adaptations
Object Abstraction
MASH
Space Dynamic/Static
Dimensions
Homogeneous/Heterogeneous
Object Composite
Atomic
Document
Relationship Association
Quality
Scope
Explicit
Implicit
Absolute
Relative
Extra-space
Intra-space
Adaptive mechanisms can
hide or show different objects
within the presentation in
accordance to the user’s preferences.
Conflict between multiple models can be resolved to fit relevant aspects of the use context, such as activity and situation.
WARP is a first implementation of a MASH-based system.
WARP can discover the implicit structure of a
document as defined by the spatial relationships
between objects.
The spatial structure provides a useful
definition of context that facilitates the adaptation
of the document.MASH