T40/edge-aui-framework-data

Dataset Card: Edge-Native Adaptive UI Behavioral Logs

This repository stores aggregated behavioral interaction logs in raw and processed forms, primarily in the .parquet data storage format. These logs support research on edge-native adaptive user interfaces.

1. Hosted Data

The repository hosts microtensor parquet files derived from public research datasets and runtime experiment sessions:

• —AdSERP Search and Interaction Logs (Arapakis et al., 2025).
• —High-Volume Trajectories (Mendeley Mouse Dynamics, 2026).
• —Structured Human-Machine Interaction Logs (Carrera-Rivera et al., 2023).
• —Client-Side Action Paths (Ou et al., 2021).
• —Continuous Kinematics (Leiva and Arapakis, 2020).

Interim & Sequence Datasets:

• —Canonical Traces (`interim/traces/canonical_traces.parquet`): Ingested browser experiment telemetry traces (schemaVersion 1.1.0) with standard and extended provenance fields (sessionId, experimentId, conditionId, taskId, windowId, sourceEventIds, preprocessingVersion, featureSchemaVersion).
• —MicroTensor Sequences (`interim/sequences.parquet`): Model-ready sequential interaction tensors ($T=8$, $D=18$) assembled with strict session and task boundary isolation and lookahead window alignment for TargetInterventionHead training.

Raw Scripted Traces:

• —Verified Testbed Traces (`raw/scripted/`): 18 verified browser experiment traces (schemaVersion 1.1.0) recorded from automated browser sessions across baseline and adaptive conditions, complete with trace validation and window attribution verification (manifest.json).

Processed Datasets:

• —Target Intervention Dataset (`processed/v1.0.0/target_intervention_dataset.parquet`, `processed/target_intervention_dataset.parquet`): Target-domain dataset with 289 examples pairing sequential MicroTensors ($T=8$, $D=18$) and canonical UIContext vectors ($\mathbb{R}^6$) with 5-class intervention labels (no_op, highlight_primary_action, simplify_options, expand_tooltip, offer_assistance) generated via Scripted Intervention Label Policy v1.0.0 (ADR-013). Accompanied by provenance-versioned manifest.json detailing SHA-256 trace digests, class distributions, and session-bounded train/val/test splits.

2. References

For full details on the data preprocessing pipeline, MicroTensor feature schema, and target outcome taxonomy, refer to the following repositories:

• —Model Preparation Repository: taofeeqhamzat/edge-aui-model-preparation
• —Framework Repository: taofeeqhamzat/edge-aui

Please refer to the source repositories for detailed technical specifications and training guidelines.