hugosousa/TemporalGamesLevels

该数据集包含两个配置：clean和raw。每个配置包含以下特征：id（字符串类型）、context（字符串类型）、entities（字符串序列）、timeline（包含relation、source和target的列表）、level_0、level_1和level_2（均为字符串类型）。数据集分为训练集和测试集，训练集和测试集的大小和示例数量在clean和raw配置中有所不同。clean配置的训练集包含86637个示例，测试集包含2400个示例；raw配置的训练集包含96310个示例，测试集包含2700个示例。数据集可能用于涉及上下文、实体和时间线关系的自然语言处理任务。

The dataset contains two configurations: clean and raw. Each configuration includes the following features: id (string type), context (string type), entities (sequence of strings), timeline (a list containing relation, source, and target), level_0, level_1, and level_2 (all string types). The dataset is divided into training and test sets, with different sizes and numbers of examples in the clean and raw configurations. The clean configurations training set contains 86,637 examples, and the test set contains 2,400 examples; the raw configurations training set contains 96,310 examples, and the test set contains 2,700 examples. The dataset is likely intended for natural language processing tasks involving context, entities, and timeline relationships.