Biowikinet: A Multilingual Network Dataset Revealing Biodiversity Knowledge Gaps

BioWikiNet is a multilingual dataset that links Wikipedia articles on biological taxa to the GBIF Backbone Taxonomy, enabling cross-lingual analysis of biodiversity knowledge representation. Based on the January 2025 Wikipedia snapshot, it covers 23 language editions (Arabic, Bengali, Chinese, Dutch, English, French, German, Hebrew, Hindi, Indonesian, Italian, Japanese, Korean, Malay, Persian, Polish, Portuguese, Russian, Spanish, Swahili, Thai, Ukrainian, and Vietnamese), spanning the world's most spoken languages alongside major languages of biodiversity hotspot regions. It comprises 3,076,527 articles linked to 1,474,707 taxonomic entities of the GBIF Backbone Taxonomy.The dataset includes:A core metadata file (biowikinet_core.csv) containing article-level metrics: pageviews (2024), edit history, link statistics, and derived indices including SCI (Species Connectivity Index), core index, and excess focus. Articles may appear in multiple rows when matched to several accepted or synonym GBIF taxa (3,426,227 rows in total).- A structured JSON file of article nodes (nodes.json), each annotated with its Wikidata entity ID, taxonomic rank, GBIF ID, language, and key visibility metrics.A nested JSON file of 16,393,308 taxonomic hyperlinks (edges.json) to other biological taxa, structured as adjacency lists.This dataset is derived from:Wikipedia content (primarily CC-BY-SA 4.0, with some older content under CC-BY-SA 3.0*) — article metadata and link structuresWikidata (CC0) — taxonomic entity identifiersGBIF Backbone Taxonomy (doi.org) accessed via GBIF.org (CC0) — taxonomic classification*Wikipedia transitioned to CC-BY-SA 4.0 in June 2023. Content edited after this date is licensed under 4.0, while earlier contributions remain under 3.0 (both are ShareAlike-compatible).Applications: The dataset supports research in conservation culturomics, biodiversity informatics, multilingual knowledge equity, and AI/NLP applications. It enables detection of taxonomic knowledge gaps, cultural biases in species representation, and language-specific patterns in biodiversity engagement.BioWikiNet itself is distributed under CC-BY 4.0 license. When reusing this dataset, attribution to BioWikiNet is required. Any reuse of underlying Wikipedia content must comply with its respective CC-BY-SA license terms.Version historyv4 (2026-05): Expanded language coverage from 11 to 23 editions, adding Bengali, Dutch, Hebrew, Italian, Korean, Malay, Persian, Polish, Swahili, Thai, Ukrainian, and Vietnamese. Articles grew from 1,266,215 to 3,076,527 (2.4×), GBIF taxa from 751,843 to 1,474,707 (2.0×), and inter-article links from 6,955,289 to 16,393,308 (2.4×). Source extraction migrated from NDJSON to Wikimedia Parquet dumps (2025-01 snapshot); article-level JSON outputs are now generated via chunked writers. Pageviews are pre-aggregated to mitigate bot/stub-page noise.v3 (2025): Initial release of 11 major language editions (Arabic, Chinese, English, French, German, Hindi, Indonesian, Japanese, Portuguese, Russian, Spanish).When citing this dataset, please use: Shinya Uryu (2026). BioWikiNet: A Multilingual Network Dataset Revealing Biodiversity Knowledge Gaps. Figshare. doi.org