Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Toward Strategy Identification and Subtask Decomposition In Task Exploration

Tom Odem

TL;DR

The paper addresses enabling anticipatory human–machine coordination by automatically extracting key task strategies and meaningful, transferable subtasks from user-run time-series data. It introduces a task explorer pipeline that combines term-frequency analysis, factor analysis, hierarchical clustering, and string-distance based composition to identify global (BoT) and local (Echo) strategies and to hierarchically encode runs with subtasks, supported by a GUI for review. Key contributions include automatic discovery of hierarchical subtasks via collocations and PMI, two-level strategy identification, hierarchical run encoding, and a practical visualization tool that can adapt to action-based data beyond the presented cyber security setting. The work advances implicit coordination in aHMI with potential applications for adaptive interfaces and intelligent tutoring, and sets the stage for future enhancements in parameter generalization and predictive task understanding across domains.

Abstract

This research builds on work in anticipatory human-machine interaction, a subfield of human-machine interaction where machines can facilitate advantageous interactions by anticipating a user's future state. The aim of this research is to further a machine's understanding of user knowledge, skill, and behavior in pursuit of implicit coordination. A task explorer pipeline was developed that uses clustering techniques, paired with factor analysis and string edit distance, to automatically identify key global and local strategies that are used to complete tasks. Global strategies identify generalized sets of actions used to complete tasks, while local strategies identify sequences that used those sets of actions in a similar composition. Additionally, meaningful subtasks of various lengths are identified within the tasks. The task explorer pipeline was able to automatically identify key strategies used to complete tasks and encode user runs with hierarchical subtask structures. In addition, a Task Explorer application was developed to easily review pipeline results. The task explorer pipeline can be easily modified to any action-based time-series data and the identified strategies and subtasks help to inform humans and machines on user knowledge, skill, and behavior.

Toward Strategy Identification and Subtask Decomposition In Task Exploration

TL;DR

The paper addresses enabling anticipatory human–machine coordination by automatically extracting key task strategies and meaningful, transferable subtasks from user-run time-series data. It introduces a task explorer pipeline that combines term-frequency analysis, factor analysis, hierarchical clustering, and string-distance based composition to identify global (BoT) and local (Echo) strategies and to hierarchically encode runs with subtasks, supported by a GUI for review. Key contributions include automatic discovery of hierarchical subtasks via collocations and PMI, two-level strategy identification, hierarchical run encoding, and a practical visualization tool that can adapt to action-based data beyond the presented cyber security setting. The work advances implicit coordination in aHMI with potential applications for adaptive interfaces and intelligent tutoring, and sets the stage for future enhancements in parameter generalization and predictive task understanding across domains.

Abstract

This research builds on work in anticipatory human-machine interaction, a subfield of human-machine interaction where machines can facilitate advantageous interactions by anticipating a user's future state. The aim of this research is to further a machine's understanding of user knowledge, skill, and behavior in pursuit of implicit coordination. A task explorer pipeline was developed that uses clustering techniques, paired with factor analysis and string edit distance, to automatically identify key global and local strategies that are used to complete tasks. Global strategies identify generalized sets of actions used to complete tasks, while local strategies identify sequences that used those sets of actions in a similar composition. Additionally, meaningful subtasks of various lengths are identified within the tasks. The task explorer pipeline was able to automatically identify key strategies used to complete tasks and encode user runs with hierarchical subtask structures. In addition, a Task Explorer application was developed to easily review pipeline results. The task explorer pipeline can be easily modified to any action-based time-series data and the identified strategies and subtasks help to inform humans and machines on user knowledge, skill, and behavior.

Paper Structure

This paper contains 49 sections, 2 equations, 15 figures, 10 tables.

Table of Contents

  1. Introduction
  2. Human-Machine Interaction
  3. Anticipatory Human-Machine Interaction
  4. Types Of Anticipation
  5. Theory Of Mind
  6. Gaps & Challenges In aHMI
  7. Research Objective
  8. Literature Review
  9. Subtask Identification
  10. Strategy Detection
  11. Summary
  12. Methods
  13. Capture The Flag Dataset
  14. Input Data Format
  15. Term Frequency Vectorization
  16. ...and 34 more sections

Figures (15)

  • Figure 1: Example Task's k Found Via The Kneedle Algorithm
  • Figure 2: Example Task's Two Dimensional Representation Of k = 8 Clusters
  • Figure 3: Dendrogram Produced From Hiearchical Clusters In Example Task
  • Figure 4: The $\epsilon$ Elbow Found Via The Kneedle Algorithm For Compositional Clustering In Term Frequency Cluster 0
  • Figure 5: Hierarchical Encoding Diagrams For Some Runs In Compositional Clusters From Term Frequency Cluster 6
  • ...and 10 more figures