DisasterLex DuckDB

DuckDB binary used as the relational backend for the DisasterLex disaster-analytics case study. 36 geospatial tables on the H3 resolution-8 hexagonal grid (827,648 cells per table, ~0.74 km^2 per cell), built from public-domain U.S. federal data sources: FEMA National Risk Index (per-hex hazard scores for flood, hurricane, tornado, wildfire), US Census + ACS (population and exposure), CDC/ATSDR Social Vulnerability Index, FEMA NRI Community Resilience Index, and Homeland Infrastructure Foundation-Level Data (hospitals, fire stations, shelters, power plants). All tables joined on a shared hex_id primary key plus a county/state/ZIP crosswalk.

This deposit is anonymous for EMNLP 2026 double-blind review. Public release upon paper acceptance.

SHA-256: 58a9eb5acfed3bcd0b0e1e77048c205e415d1d7bf59630016ac72bfd86a52138

本数据集采用DuckDB二进制文件作为关系型后端，用于DisasterLex灾害分析案例研究。数据集包含36个基于H3分辨率8级六边形网格构建的地理空间数据表（每张表含827648个网格单元，单个单元面积约0.74平方千米），所有数据均取自公有领域的美国联邦数据源：联邦紧急事务管理局（Federal Emergency Management Agency, FEMA）国家风险指数（各六边形网格对应的洪水、飓风、龙卷风、野火灾害风险得分）、美国人口普查局（U.S. Census）与美国社区调查（American Community Survey, ACS）数据（人口与暴露度）、美国疾病控制与预防中心/毒物与疾病登记署（CDC/ATSDR）社会脆弱性指数、FEMA国家风险指数社区韧性指数，以及国土基础设施基础级数据（Homeland Infrastructure Foundation-Level Data，包含医院、消防站、避难所、发电厂相关信息）。所有数据表均通过共享的hex_id主键以及县/州/邮政编码跨映射表完成关联。
本数据集以匿名形式提交，用于EMNLP 2026双盲审稿流程，待论文正式录用后将公开发布。
SHA-256哈希值：58a9eb5acfed3bcd0b0e1e77048c205e415d1d7bf59630016ac72bfd86a52138