Dataset on material composition and recyclability constraints of food packaging designs in global B2B supply chains

Overview The repository provides design-level evidence on material composition, laminate structures, component compatibility, and recycling incompatibility in food packaging. The data are derived from large-scale scraping and structured parsing of global business-to-business (B2B) packaging listings on Alibaba.com and are linked to food product categories defined under the EU Combined Nomenclature (CN). The materials support: reproducibility of the results reported in the paper, reuse in material flow analysis (MFA), systematic assessment of Design-for-Recycling (DfR) alignment, further research on structural lock-in in packaging systems. Data scope Food product coverage2,397 EU Combined Nomenclature (CN) food codes(Chapters 02–23; excluding non-food Chapters 05, 06, 13, and 14) Packaging design data41,000 packaging listings scraped from Alibaba.com(up to 500 listings per packaging keyword; 82 keywords) Supply-chain contextGlobal B2B marketplace listings from Original Equipment Manufacturers (OEM) and Original Design Manufacturers (ODM) Repository structure food_packaging_design_data.zip││── code/│ ├── 1_keyword_mapping.py│ ├── 2_collect_urls.py│ ├── 3_fetch_attributes.py│ └── 4_result_parse.py││── data_raw/│ ├── flat_merged_attributes_urls_packaging_<keyword>_500_<timestamp>.csv│ └── (82 files; one per packaging keyword)││── data_processed/│ │── all_listings/│ │ ├── listings_<keyword>.csv (82 files)│ ││ │── all_offerings/│ │ ├── offerings_<keyword>.csv (82 files)│ ││ └── all_metadata/│ ├── metadata_<keyword>.csv (82 files)││── readouts/ ├── df_mapping_rules.csv ├── all_shares_per_keyword.xlsx └── extended_share_results.xlsx 1. code/ — analysis pipeline Python scripts implementing the full workflow: 1_keyword_mapping.pyMaps CN food product codes to packaging-relevant search keywords using a transparent, rule-based system. 2_collect_urls.pyRetrieves Alibaba listing URLs for each packaging keyword. 3_fetch_attributes.pyScrapes structured attribute tables from each listing. 4_result_parse.pyNormalizes material information, reconstructs laminate structures and components, computes offering- and sales-weighted share indicators, and generates final readouts. Scripts are designed to be executed sequentially. 2. data_raw/ — raw scraped listing-level data Contains 82 CSV files of the form: flat_merged_attributes_urls_packaging_<keyword>_500_<timestamp>.csv Each file corresponds to one packaging keyword and contains up to 500 Alibaba listings. Column groups include: Script-defined control and traceability fields(count, page, url, title, highlighted_tokens, highlight_count,url_collecting_time, attribute_fetch_time, sold) Raw Alibaba attribute fieldsMaterial descriptions, packaging form descriptors, component information, and commercial metadata as provided by sellers. These files preserve the original scraped evidence and support transparency and reprocessing. 3. data_processed/ — harmonized analytical datasets a) all_listings/ Cleaned, listing-level datasets with harmonized material fields. Key columns include: material_parsed, options_str layer_counts, is_laminate has_component, component_materials component_match_body w_offering, w_sales b) all_offerings/ Expanded offering-level datasets, where each listing may generate multiple offerings reflecting alternative material configurations. Additional analytical fields include: layer_class NR_sub_share_type nr_flag These data form the analytical basis for all reported material-share indicators. c) all_metadata/ Per-keyword metadata reporting: number of listings, number of offerings, parsed-offering counts, sales-data availability and coverage. 4. readouts/ — final analytical outputs a) df_mapping_rules.csv Rule-based mapping table linking CN food codes to packaging keywords using regular expressions.This file documents the semantic and functional logic used to translate trade classifications into packaging search categories. b) all_shares_per_keyword.xlsx Offering-weighted share results for each of the 82 packaging keywords, including: packaging-composition share, layer-complexity share, component-presence share, recycling-incompatibility share. Each sheet corresponds to one packaging keyword. c) extended_share_results.xlsx Cross-packaging aggregated results structured by: share type (composition, layers, components, incompatibility), weighting scheme (offering-weighted, sales-weighted). This file underlies the figures and tables reported in the paper and Supplementary Information. Relation to the paper Supplementary Dataset (journal)Contains the main aggregated results ('df_mapping_rules.csv', 'all_shares_per_keyword.xlsx', and 'extended_share_results.xlsx') used for interpretation. This Zenodo repositoryProvides the full underlying data, intermediate datasets, and code required to reproduce or extend the analysis. Intended reuse The dataset is suitable for: material flow analysis (MFA), Design-for-Recycling (DfR) evaluation, packaging policy analysis, comparative studies of global packaging design practices. Citation Please cite the associated article when using this dataset.A Zenodo DOI is provided for direct citation of this repository.