Human-only and Human-AI Collaborative Decision-Making for Higher Education

Overview

The dataset contains 40 XLSX files derived from a study-tour digital operation sandbox experiment conducted under two collaboration conditions:

Human-only collaboration (Human)

Human–AI Collaboration (HAC)

Each condition contains 20 files, resulting in a total of 40 files.

The dataset records information dependency and task co-occurrence relationships among operational modules during multi-round decision-making processes.

File Structure

A. Human-only Dataset (20 files)

1. Individual Round Matrices (18 files)

These files represent the original matrices collected from three independent simulation runs.

File naming format:

Human-X-Y

where:

X = simulation run number (1–3)

Y = decision-making round (1–6)

Examples:

Human-1-1Human-1-2...Human-3-6

A total of:

3 simulation runs × 6 rounds = 18 files

2. Aggregated Network Files (2 files)

Human-combination

This file contains:

Aggregated matrices for each simulation run

Overall aggregation across all runs and rounds

Specifically:

Run 1: Rounds 1–6 combined

Run 2: Rounds 1–6 combined

Run 3: Rounds 1–6 combined

Grand network: all rounds from all runs combined

Human-turns' combination

This file contains cross-run aggregations by round.

For each round, matrices from the three simulation runs are summed:

Dataset

Contents

Round 1

Run1-R1 + Run2-R1 + Run3-R1

Round 2

Run1-R2 + Run2-R2 + Run3-R2

Round 3

Run1-R3 + Run2-R3 + Run3-R3

Round 4

Run1-R4 + Run2-R4 + Run3-R4

Round 5

Run1-R5 + Run2-R5 + Run3-R5

Round 6

Run1-R6 + Run2-R6 + Run3-R6

This file is intended for analyzing structural evolution across decision-making rounds.

B. Human–AI Collaboration Dataset (20 files)

1. Individual Round Matrices (18 files)

File naming format:

Human-AI-X-Y

where:

X = simulation run number (1–3)

Y = decision-making round (1–6)

Examples:

Human-AI-1-1Human-AI-1-2...Human-AI-3-6

Total:

3 simulation runs × 6 rounds = 18 files

2. Aggregated Network Files (2 files)

Human-AI combination

Contains:

Aggregated matrices for Run 1

Aggregated matrices for Run 2

Aggregated matrices for Run 3

Overall aggregation across all runs and rounds

Specifically:

Run 1: Rounds 1–6 combined

Run 2: Rounds 1–6 combined

Run 3: Rounds 1–6 combined

Grand network: all rounds from all runs combined

Human-AI-turns' combination

Contains cross-run aggregation by round.

For each round:

Dataset

Contents

Round 1

Run1-R1 + Run2-R1 + Run3-R1

Round 2

Run1-R2 + Run2-R2 + Run3-R2

Round 3

Run1-R3 + Run2-R3 + Run3-R3

Round 4

Run1-R4 + Run2-R4 + Run3-R4

Round 5

Run1-R5 + Run2-R5 + Run3-R5

Round 6

Run1-R6 + Run2-R6 + Run3-R6

These files support temporal comparisons of decision-making structures under Human–AI collaboration.

Matrix Interpretation

In all files:

Rows and columns represent task modules (sub-modules).

Cell values indicate information-reference frequency between task modules.

Larger values indicate stronger dependency or co-occurrence relationships.

Matrices can be directly imported into UCINET, Gephi, Pajek, R (igraph), or Python (NetworkX) for network analysis.

Summary of Dataset Composition

Dataset Type

Number of Files

Human-only original matrices

18

Human-only aggregated files

2

Human–AI original matrices

18

Human–AI aggregated files

2

Total

40

Note: Although the sandbox platform consists of seven operational rounds, only the first six rounds involve participant decision-making. The seventh round is automatically generated by the system and is therefore not included in the released dataset.