overthelex/ua-court-citation-graph
收藏资源简介:
这是一个大规模共引图数据集,基于乌克兰法院判决统一国家登记册(ЄДРСР / EDRSR)中的9950万份法院判决构建。数据集包含乌克兰法律条款(如法律条文)的共引网络,时间跨度为2003年至2026年。当两个法律条款在同一份法院判决中被共同引用时,它们之间会建立连接,边的权重表示共同引用该对条款的判决数量。数据集通过正则表达式从判决文本中提取了3.45亿条引用,并过滤出权重≥10的共引边,最终包含约232.8万条边和5万个唯一法律条款。数据集文件包括:共引边文件(cocitation_edges),记录条款对及其权重;条款引用统计文件(article_citation_stats),列出最常被引用的条款及其引用次数;条款中心性文件(article_centrality),提供图的中心性度量(如度、PageRank);社区标签文件(community_labels),展示基于时间段的社区检测结果。该数据集适用于图机器学习、文本分类、法律网络分析等任务,可用于研究法律引用模式、社区结构和预测分析。关键发现包括网络具有幂律度分布(指数为2.17)、社区结构与法律领域(如刑事、民事法)对齐,以及时间稳定性高。数据集遵循CC-BY-NC-SA-4.0许可,学术和研究使用免费。
A large-scale co-citation graph derived from 99.5 million Ukrainian court decisions in the Unified State Register of Court Decisions (ЄДРСР / EDRSR). The dataset contains the co-citation network of Ukrainian legal provisions (articles) from 2003 to 2026, where two provisions are linked if they are cited together in the same court decision, with edge weights reflecting the number of decisions co-citing the pair. It includes extracted citations via regex patterns, filtered to edges with weight >= 10, resulting in approximately 2.33 million edges and 50,000 unique articles. Files include: co-citation edges (cocitation_edges) with pairwise article connections and weights; article citation statistics (article_citation_stats) for top-cited provisions; article centrality measures (article_centrality) such as degree and PageRank; and community labels (community_labels) from Louvain detection across time periods. The dataset is suited for graph ML, text classification, and legal network analysis, with key findings like power-law degree distribution (alpha=2.17), community alignment to legal domains, and high temporal stability. Licensed under CC-BY-NC-SA-4.0 for academic and research use.




