Salima Hassaine, Karim Dhambri, Houari Sahraoui, Pierre Poulin

1
Generating Visualization-based Analysis
Scenarios from Maintenance Task Descriptions

• Salima Hassaine, Karim Dhambri, Houari Sahraoui,
  Pierre Poulin
• DIRO, Université de Montréal

2
Motivation

• Many maintenance tasks difficult to automate
• Multi-criteria decision making
• Feature location (traceability), anomaly
  detection, aspect mining, component
  identification, etc.
• Lack of contextual information
• Feature location, quality assessment, anomaly
  detection, evolution analysis, reverse
  engineering, etc.
• Complexity/scalability
• Feature location, quality assessment, anomaly
  detection, evolution analysis, reverse
  engineering, etc.

3
Motivations

• Interactive visualization
• Maintenance task Set of reasoning and
  calculation modules
• Set of automated modules AMs (explicit knowledge)
• Human analyst module HM
• Visualization Interface between AMs and HM

4
Motivations

• How to us visualization tools?

5
Motivations

• How to us visualization tools?
• Creating views
• Mapping data to graphical objects and attributes
• Different degrees of freedom
• Fully defined by the tool designer
• Customized by parameter setting
• More flexibility
• Interacting with views
• Mapping data-analysis operations to sequences of
  interactions

6
Approach Overview
Interactive Visualization Task Model
AT to IVT Transformation Mechanism
Analysis Task Model
Edition
Execution
Conformance
AT to IVT Transformation Session
Analysis Task Model
Interactive Visual Task
Customization
Tool Specification
7
Analysis Task Model
8
Example

• Blob Detection
• Task description

9
Interactive-Visualization Task Model

• Tool specification
• Example of VERSO

10
Interactive-Visualization Task Model
11
AT to IVT Transformation

• Defining views (case of parameter setting)
• Map data into graphical attributes
• Constraint solving problem

12
AT to IVT Transformation

• Defining interaction scenarios
• Mapping operators into interactors (two steps)

13
Example

• Blob Detection
• Interaction scenario

14
Example

• Blob Detection

15
Conclusion

• A preliminary work on how to map a maintenance
  task to an interaction scenario with a
  visualization tool
• Limitations Future work
• Relationship between data and operation mappings
• Data-distribution impact on the mapping
• Tool-nature impact on the mapping
• Case of multiple views
• Experimentation with other visualization tools