factnet_factstatements

# FactStatement Dataset ## Overview FactStatement is the foundational layer of FactNet, a cross-lingual, multi-layered fact knowledge graph. FactStatements are language-neutral, atomic fact units directly mapped from Wikidata statements, forming the core building blocks of the knowledge graph. + Paper: https://arxiv.org/abs/2602.03417 + Github: https://github.com/yl-shen/factnet + Dataset: https://huggingface.co/collections/openbmb/factnet ## Dataset Format The dataset contains parquet files with the following key fields: - `core_id`: Unique identifier for the fact statement - `subject_qid`: Wikidata QID of the subject entity - `property_pid`: Wikidata PID of the property - `value`: Raw value from Wikidata - `value_qid`: Entity QID if the value is an entity (otherwise null) - `normalized_value`: Standardized representation of the value - `claim_hash`: Standardized hash of subject|property|normalized_value|qualifiers - `qualifiers`: Qualifier information (time, location, etc.) - `references`: Source information - `rank`: Wikidata rank (preferred, normal, deprecated) - `confidence`: Computed confidence score ## Usage FactStatements are designed to be language-neutral representations of facts that can be linguistically realized through the FactSense layer and semantically grouped through the FactSynset layer. ## License This dataset is derived from Wikidata and is available under the CC0 license. ## Citation ``` @article{shen2026factnet, title={FactNet: A Billion-Scale Knowledge Graph for Multilingual Factual Grounding}, author={Shen, Yingli and Lai, Wen and Zhou, Jie and Zhang, Xueren and Wang, Yudong and Luo, Kangyang and Wang, Shuo and Gao, Ge and Fraser, Alexander and Sun, Maosong}, journal={arXiv preprint arXiv:2602.03417}, year={2026} } ```