ProPACT Hybrid-AI feedback system with five feedback tools and hierarchical trigger policy¶

Design · All designs
Evidence · no claims cited

Description¶

The feedback system is a Hybrid-AI framework that "combines data-driven forecasting with rule-based pedagogical decision-making". An XGBoost model predicts JVA, JME, and ME over a 30-second horizon; predictions are discretized against each participant's resting baseline using a ±2SD criterion and mapped against a desired collaboration matrix to predefined trigger conditions. Five feedback tools are integrated: GitHub Copilot autocomplete, dual text selection, a gaze-awareness tool, a dialogue prompt, and a last-resort task-based hint, selected via a top-down hierarchical policy prioritizing minimal intervention.

Design Implications¶

Context¶

Requirements¶

A desired collaboration matrix and predefined trigger conditions (e.g., gaze-awareness triggered when JVA = Low, dialogue prompt when JME = Low, task-based hint when both MEs = High)
Per-participant resting baselines for the ±2SD discretization of predicted states

Constraints¶

The task-based hint is used only as a last-resort intervention, triggered when less intrusive scaffolds fail and both collaborators exhibit sustained extreme mental-effort levels

Target Learners¶

Undergraduate or master's computer science and engineering students pair programming on debugging tasks

Learning Goals¶

Restoring joint attention and cognitive alignment during collaborative debugging

Claims¶

Propact Dyadic Forecast Driven Adaptivity Framework [+M]

Hybrid micro-level personalization framework combining learner-driven preferences with system-driven Bloom's Taxonomy cognitive assessment

Examples¶

-

Key Sources¶

Anahita Golrang, Kshitij Sharma, Simon Dehaen, Olga Viberg. (2026). ProPACT: A Proactive AI-Driven Adaptive Collaborative Tutor for Pair Programming. https://arxiv.org/abs/2605.02703

Page metadata

Field	Value
`type`	`design`
`id`	`propact-hybrid-ai-feedback-system-five-tools`
`title`	`ProPACT Hybrid-AI feedback system with five feedback tools and hierarchical trigger policy`
`description`	`The feedback system is a Hybrid-AI framework that "combines data-driven forecasting with rule-based pedagogical decision-making".`
`status`	`review`
`generated`	by `process:settle-candidates` on `2026-10-11`
`trust`	unverified — no one has confirmed this page's content
`sources`	1 — `anahita-golrang-2026`

Frontmatter is the page's machine-readable record — see the Schema & Guide.