Ishaank18/screenplay-features

该数据集包含从MENSA数据集的电影剧本中提取的913个语言特征，用于剧本场景显著性检测。这些特征分为24个特征组，涵盖了语言、叙事和话语分析的各个方面。数据集分为训练集（117,503个样本，172.9 MB）、验证集（8,052个样本，16.1 MB）和测试集（8,156个样本，16.1 MB），总共有133,711个样本，大小为140.1 MB。每个parquet文件包含电影ID、场景索引、显著性标签（0表示非显著场景，1表示显著场景）和各种语言/叙事特征。特征分为基础特征（基本语言统计和结构位置特征）、GenreClassifier特征（包括基本语言指标、角色多样性指标、具体性分数、对话特定特征等）、叙事特征（角色发展指标、情节进展指标、叙事结构特征、情感弧特征）和语言特征（N-gram多样性指标、修辞结构理论特征、BERT-based surprisal分数等）四大类。

This dataset contains 913 linguistic features extracted from movie screenplays in the MENSA dataset for screenplay scene salience detection. Features are organized into 24 feature groups covering various aspects of linguistic, narrative, and discourse analysis. The dataset is divided into training set (117,503 samples, 172.9 MB), validation set (8,052 samples, 16.1 MB), and test set (8,156 samples, 16.1 MB), totaling 133,711 samples with a size of 140.1 MB. Each parquet file contains movie ID, scene index, salience label (0 for non-salient scene, 1 for salient scene), and various linguistic/narrative features. Features are categorized into base features (basic linguistic statistics and structural position features), GenreClassifier features (including basic linguistic metrics, character diversity metrics, concreteness scores, dialogue-specific features, etc.), narrative features (character development metrics, plot progression indicators, narrative structure features, emotional arc features), and linguistic features (N-gram diversity metrics, Rhetorical Structure Theory features, BERT-based surprisal scores, etc.).