DataPatent/humanoid-robotics-lakehouse

Humanoid Robotics & Bipedal Systems: Scientific Literature Lakehouse

Course: Innovation Management · Politecnico di Torino
Domain: Humanoid Robotics & Dynamic Bipedal Systems
Repository Owner: MatteoAld
Canonical Data Artifact: data/scientific_literature/train.parquet

---

1. Executive Summary & Purpose

This repository provides an empirical, curated scientific literature lakehouse for humanoid robotics and dynamic bipedal systems. It was developed as the scientific grounding foundation for technology landscaping, unsupervised taxonomy discovery, and patent classification bridging (CPC).

By harvesting high-impact peer-reviewed literature ($FWCI \ge 1.5$, elite review hubs, and core IEEE ICRA/IROS conference proceedings), this corpus allows researchers to bypass patent legalese and uncover the fundamental mechanical, control, and kinematic vocabulary governing dynamic legged robots.

---

2. Dataset Architecture & Schema

The dataset is stored in columnar Snappy-compressed Apache Parquet format (data/scientific_literature/train.parquet) with the following strict schema:

Column Name	Type	Description	Analytical Purpose
id	string	Canonical OpenAlex Work URI	Primary key for deduplication
doi	string	Digital Object Identifier	Exact cross-referencing and author attribution
title	string	Publication title	Lexical analysis and semantic embedding
publication_year	int32	Year of publication	Longitudinal CAGR and timeline tracking
publication_date	string	ISO publication date	Priority dating benchmarks
cited_by_count	int32	Raw citation count	Global academic impact baseline
fwci	float32	Field-Weighted Citation Impact	Cohort-normalized bibliometric quality sieve ($FWCI \ge 1.5$)
is_in_top_10_percent	boolean	Top 10% citation rank	High-impact article filtering
primary_topic_id	string	Immutable OpenAlex Topic URI	Permanent cluster identification
primary_topic_name	string	Leiden citation topic name	Primary technology pillar grouping
subfield_name	string	Scopus/ASJC subfield	High-level disciplinary classification
field_name	string	Broad academic field	Boundary bounding (Engineering, CS)
venue_name	string	Journal or conference venue	Venue prestige filtering
keywords	list[string]	Peer-reviewed / ML phrases	Vocabulary candidate pooling
concepts	list[string]	Knowledge graph entities (Level $\ge 1$)	High-level taxonomy verification
abstract_text	string	Reconstructed continuous text	Dense neural embeddings (all-MiniLM-L6-v2, specter)

---

3. Data Provenance & Legal Governance

3.1. OpenAlex Academic Data (Public Domain Dedication)

• Harvesting Engine: Harvested via the OpenAlex REST API through the polite pool.
• Licensing: OpenAlex metadata is explicitly dedicated to the public domain under the Creative Commons CC0 1.0 Universal (CC0 1.0) Public Domain Dedication. Anyone may copy, modify, distribute, or build upon the data without restriction.
• Abstracts: Reconstructed from the native abstract_inverted_index for computational linguistics and bibliometric research.

3.2. Patent Data & Classification (EPO PATSTAT)

• Patent Disclosures: All patent references, classifications (CPC/IPC), and family metrics mapped in this project derive from public patent gazettes published by the European Patent Office (EPO) and worldwide offices.
• Legal Status: Patent documents are official statutory public disclosures. This research dataset is curated strictly for non-commercial academic research and technology landscaping under EU Directive 2019/790 (Articles 3 & 4: Text and Data Mining for Scientific Research).
• Disclaimer: This dataset is an academic research derivative and does not constitute an official legal patent register. For legal validity, consult official registers (e.g., the EPO European Patent Register or Espacenet).

3.3. Proprietary Boundaries

• This public repository contains solely publicly available academic and patent metadata. No confidential client deliverables, internal roadmaps, or proprietary business logic are included.

---

4. Citation & Attribution

If you utilize this dataset in academic research or technology landscaping, please cite the primary data creators:

OpenAlex Citation

@article{priem2022openalex,
  title={OpenAlex: A fully-open index of scholarly works, authors, venues, institutions, and concepts},
  author={Priem, Jason and Piwowar, Heather and Orr, Richard},
  journal={arXiv preprint arXiv:2205.01833},
  year={2022}
}

European Patent Office (EPO PATSTAT) Citation

@misc{epo_patstat,
  title = {PATSTAT Global: Worldwide Patent Statistical Database},
  author = {{European Patent Office}},
  year = {2026},
  publisher = {EPO},
  url = {https://www.epo.org/searching-for-patents/business/patstat.html}
}

---

5. Notice & Takedown Policy

If you are a copyright holder or author and wish to request the exclusion of your publication abstract or metadata from this academic research benchmark, please open an Issue on this Hugging Face repository.