dimbaby/FinReflectKG

FinReflectKG是一个大规模的金融知识图谱数据集，从2014年至2024年标准普尔500公司的10-K SEC文件中提取，提供了金融关系、实体和时间信息的结构化表示。数据集包含1750万条规范化的三元组，每个三元组以（实体、关系、目标）的格式表示结构化事实，并带有时间边界和丰富的上下文信息。它覆盖了743家公司，时间跨度为10多年，包括完整的文本上下文、规范化的实体和关系，以及开始/结束日期等时间信息。数据集适用于金融NLP、知识图谱构建、时间分析、风险评估和研究应用。

FinReflectKG is a large-scale financial knowledge graph dataset that provides structured representations of financial relationships, entities, and temporal information extracted from S&P 500 companies 10-K SEC filings spanning 2014-2024. It contains 17.51 million normalized triplets in the format (entity, relationship, target) with temporal bounds and rich contextual information. The dataset covers 743 companies over 10+ years, includes full text context, normalized entities and relationships, and temporal information with start/end dates. It is suitable for financial NLP, knowledge graph construction, temporal analysis, risk assessment, and research applications.