Helping Model User Interest from User Interaction

In many visualizations, user interaction is represented as an ephemeral action taken by a user that moves the system from onestate to another. After this transition is completed, the user interactions are sometimes logged, but often discarded.

For example, when a user hovers or clicks on glyphs in a scatterplot representing data items, the system may display a tooltip or a detailed view. After the appropriate response is triggered, the data representing that interaction is discarded. Given the importance of user interaction, how do we elevate user interaction in visual analytic systems to a form of data, and enable systems to compute on this data gathered from the user?

Methods in IAL can be divided into three categories: Logging methods, user model methods, and analytics methods.

Logging:

User Model:

Analytics:

Table of Contents

• Overview
• Initialization
• Logging
  • Log
  • Get Session Logs
  • Get Item Logs
  • Get Attribute Logs
  • Print Session Logs
  • Set Max Queue Size
  • Enqueue
  • Dequeue
• User Model
  • Items
    • Set Item Weight
    • Increment Item Weight
    • Get Item Score
    • Update Item Scores
  • Attributes
    • Get Attribute Weight
    • Increment Attribute Weight
    • Set Attribute Weight
    • Get Top n Points by Interaction Weights
    • Get Top n Points by Scores
    • Get n Similar Points
    • Generate Attribute Weight Vector using Similarity
    • Generate Attribute Weight Vector using Differences
  • Bias
• Analytics
  • Classification
    • K-nearest Neighbours (KNN)
  • Regression
    • Linear Regression
  • Clustering
    • K Means
    • Density-based Spatial Clustering (DBScan)
  • Dimensionality Reduction
    • PCA
  • Model Selection
  • Preprocessing

Acknowledgements