IoTutorMine: Replication Package and Benchmark for Mining Hardware Bills of Materials from IoT Tutorial Videos

This is the official replication package for the paper "IoTutorMine: A Tool for Mining Hardware Bills of Materials from IoT Tutorial Videos" submitted to the ASE 2026 Tools and Datasets Track. IoTutorMine extracts structured hardware Bills of Materials (BoMs) from IoT tutorial videos by combining YouTube transcripts with Large Language Models in a zero-shot setting. This package contains:- A benchmark of 20 manually annotated YouTube IoT tutorials (Arduino + Raspberry Pi, beginner + intermediate) with 131 ground-truth components from 16 distinct creators- 392 LLM extractions from three models: GPT-5, Gemini 3 Flash, and Claude Opus 4.7- Plain-text transcripts of all 20 videos- A component matching dictionary with 45 canonical entries and 106 variant mappings- Python evaluation scripts that reproduce the headline results of the paper (F1 = 0.933 for Gemini 3 Flash, 0.887 for GPT-5, 0.868 for Claude Opus 4.7)- Statistical analysis scripts (Wilcoxon signed-rank, Mann-Whitney U, Friedman tests) The live tool (IoTutorMine-Web) is publicly hosted at: https://ahmedbahaj.github.io/IoTutorMine/ The tool source code is archived separately at: https://doi.org/10.5281/zenodo.20134482 License: MIT for code (scripts/), Creative Commons Attribution 4.0 International (CC-BY-4.0) for data.