Dataset for D.A.T.A: Shapley-Based Fair Data Valuation for Federated IoT Process Mining with Blockchain Accountability

Replication package for the paper "D.A.T.A: Shapley-Based Fair Data Valuation for Federated IoT Process Mining with Blockchain Accountability". This dataset contains the complete source code, smart contracts, configuration, reproducibility scripts, and pre-computed experiment results for the D.A.T.A platform — a five-layer architecture integrating federated process mining, Shapley-based cooperative game-theoretic data valuation, and permissioned blockchain (Alastria Network) smart contracts for fair and auditable reward distribution in IoT federations. Contents: Python source code (src/): Federated node processing (CEP engine, pm4py-based DFG discovery), DFG aggregation, exact and TMC-Shapley value computation, quality metrics, Web3.py blockchain client, and FastAPI marketplace API. Solidity smart contracts (contracts/): ERC-20 DataToken, DataRegistry (immutable audit trail), and RewardDistributor (on-chain Shapley efficiency verification and token minting), compiled with Solidity 0.8.20 + viaIR optimiser. Deployed on Alastria T-Network (chainId 2020). Evaluation pipeline (evaluation/): Experiment runners for three benchmark IoT datasets (PAMAP2, CASAS, UCI HAR), Shapley axiom verification (efficiency, symmetry, dummy, additivity), baseline comparisons (equal-split, proportional-volume, random), convergence analysis, and correlation metrics. Pre-computed results (results/): 7 JSON files containing Shapley values, fairness metrics, convergence analysis, baseline comparisons, and blockchain transaction records (including verifiable on-chain TX hashes) for all 60 federated IoT nodes across 3 datasets. Reproducibility scripts (scripts/): Automated dataset download, experiment execution entry point, blockchain deployment (Hardhat), and on-chain reward distribution with batched transactions. Infrastructure: Docker Compose multi-node federation, Hardhat blockchain configuration, CEP rule definitions, and federation parameters. Raw datasets (PAMAP2, CASAS adlinterweave, UCI HAR) are not included due to their respective licenses but can be downloaded automatically via the provided scripts/download_datasets.py script.