VynFi/vynfi-je-network-2k

VynFi日记账网络——2,000实体集团合并边数据集是一个合成数据集，包含68,463,796条日记账网络边，由v5.31集团审计模拟引擎生成，覆盖2,000个实体及合并财务报表包。每条边代表单个日记账分录中的一个借方↔贷方关系，附带其源实体属性，并包含从生成器传递的文档链、欺诈/异常标签和公司间配对标签。数据集适用于边级欺诈检测（二元或多类分类）、账户网络/文档链分析（通过predecessor_edge_id重构文档流路径）、公司间匹配/抵消研究（标记6,814对公司间配对和13,628条消除边）以及异常注入基准测试（is_anomaly标志非欺诈异常）。数据为合成生成，不包含真实客户或语料内容，实体名称、账户和金额均为合成数据。

VynFi JE Network — 2 000-entity group consolidated edges is a synthetic dataset of 68,463,796 journal-entry network edges produced by the v5.31 group-audit simulation engine over 2,000 entities plus a consolidated financial statement bundle. Each row represents one debit↔credit edge from a single journal entry, attributed to its source entity, with document-chain, fraud/anomaly, and intercompany-pair labels carried through from the generator. The dataset is designed for edge-level fraud detection (binary or multi-class classification), account-network/document-chain analytics (reconstructing paths via predecessor_edge_id), intercompany matching/elimination research (with 6,814 IC pairs and 13,628 elimination edges), and anomaly-injection benchmarking (using is_anomaly flags). All data is procedurally generated, with no real client or corpus content; entity names, accounts, and amounts are synthetic.