Datasets:
Title stringlengths 2 118 | Genre stringclasses 5
values | Length int64 1.01k 6.51M | Confidence float64 0.09 0.99 | Text stringlengths 1.01k 6.51M | Raw Text stringlengths 22k 6.66M |
|---|---|---|---|---|---|
WHAT CHEER | Literature & Fiction | 324,907 | 0.689215 | "“And surely betweene my friends of the Bay and Plimouth, I was sorely\ntost for fourteen weeks, i(...TRUNCATED) | "The Project Gutenberg eBook of What Cheer, by Job Durfee\n\nThis eBook is for the use of anyone (...TRUNCATED) |
THE GO AHEAD BOYS | Children's Books | 233,507 | 0.224825 | "CHAPTER I—MAKING CAMP\n\n“Here is the place to put the tent, String.”\n\n“I think this spot(...TRUNCATED) | "Project Gutenberg's The Go Ahead Boys in the Island Camp, by Ross Kay\n\nThis eBook is for the u(...TRUNCATED) |
THE UPHILL CLIMB | Literature & Fiction | 294,129 | 0.192787 | "CHAPTER\n\n I \"Married! And I Don't Know Her Name!\"\n II Wanted: Information\n III One Way (...TRUNCATED) | "The Project Gutenberg eBook, The Uphill Climb, by B. M. Bower, Illustrated\nby Charles M. Russell\n(...TRUNCATED) |
THE WIDE AWAKE GIRLS IN WINSTED | Children's Books | 322,797 | 0.319238 | "child.\n\n\"Wake up, kid! Here, you've gone past your station. Wake up, I say! Gee!\nWe're running (...TRUNCATED) | "The Project Gutenberg eBook, The Wide Awake Girls in Winsted, by Katharine\nEllis Barrett, Illustra(...TRUNCATED) |
COLLOQUIES ON SOCIETY | Literature & Fiction | 228,942 | 0.627155 | "It was in 1824 that Robert Southey, then fifty years old, published \"Sir\nThomas More, or Colloqui(...TRUNCATED) | "The Project Gutenberg eBook of Sir Thomas More: or, Colloquies on the Progress and Prospects of (...TRUNCATED) |
HAYDN | Literature & Fiction | 108,320 | 0.262865 | "CHAPTER I\n\nJOSEPH HAYDN\n\nIt is, as a rule, inexpedient to begin a book with the peroration.\nCh(...TRUNCATED) | "The Project Gutenberg eBook, Haydn, by John F. Runciman\n\n\nThis eBook is for the use of anyone an(...TRUNCATED) |
MIRROR OF TASTE, APRIL 1810 | Literature & Fiction | 259,295 | 0.91765 | "CHAPTER IV.\n\nORIGIN OF COMEDY--ARISTOPHANES--DEATH OF SOCRATES.\n\nThough the term \"tragedy\" ha(...TRUNCATED) | "The Project Gutenberg EBook of The Mirror of Taste, and Dramatic Censor,\nVol. I, No. 4, April 1810(...TRUNCATED) |
THE FESTIVAL OF SPRING | Literature & Fiction | 127,032 | 0.537615 | "The current, popular spelling of Persian Names and Words has been\ngenerally adopted in the followi(...TRUNCATED) | "Project Gutenberg's The Festival of Spring, by Maulana Jal?l al-D?n R?m?\n\nThis eBook is for th(...TRUNCATED) |
THE CIVIL WAR IN AMERICA | Literature & Fiction | 466,583 | 0.365065 | "In presenting the first number of the _Modern Age_ to the public, I have\nselected the letters of M(...TRUNCATED) | "Project Gutenberg's The Civil War in America, by William Howard Russell\n\nThis eBook is for the(...TRUNCATED) |
THE JEALOUSY OF LE BARBOUILLÉ | Literature & Fiction | 21,335 | 0.758655 | "Among the small farces said to have been sketched by Molière during\nhis stay in the provinces, tw(...TRUNCATED) | "The Project Gutenberg EBook of The Jealousy of le Barbouillé, by \nJean Baptiste Poquelin de Moli(...TRUNCATED) |
End of preview. Expand in Data Studio
Dataset Card for 123Tapwater/StorySet
Dataset Description
StorySet is a curated subset of manu/project_gutenberg. This dataset contains:
- Full text of classic fiction books
- 5000+ Samples
- Cleaned text preserving chapter structure
- Genre classifications with confidence scores
Contents
The dataset contains:
- 5 core fiction genres: Mystery, Romance, Science Fiction, General Fiction, and Children's literature
- Cleaned text versions with:
- Metadata headers/footers removed
- Chapter structure preserved
- Narrative content validated
- Additional metadata including:
- Original Project Gutenberg links
- Text length metrics
- Classification confidence scores
Creation Process
Data Collection
- Source: Project Gutenberg eBooks (raw text files)
- Initial Filter: Used a pretrained transformer model to identify fiction books from generic gutenberg dataset
Preprocessing
Text Cleaning (
enhanced_preprocess):- Metadata removal (headers/footers)
- Chapter detection using regex patterns + spaCy validation
- Dialogue/narrative preservation
- Whitespace normalization
Classification:
- Fine-tuned BERT-large model (TenzinGayche/Bert-large-book-genre-classification)
- Threshold: Confidence ≥ 0.7 for genre assignments
- Fiction/Non-fiction separation using predefined labels
Quality Control
- Minimum text length: 1,000 characters
- spaCy narrative validation
- Manual spot-checking of samples
Column Descriptions
| Column | Description |
|---|---|
title |
Original book title from Gutenberg metadata |
length |
Character length of the book |
genre |
Classified genre (Mystery/Romance/Sci-Fi/etc) |
confidence |
Character count of cleaned text |
text |
Preprocessed text with chapters/narrative content preserved |
raw text |
Raw Source Text Prior to pre-processing |
Usage
from datasets import load_dataset
dataset = load_dataset("123Tapwater/StorySet")
# Example analysis
df = dataset["train"].to_pandas()
print(f"Available genres: {df['genre'].unique()}")
print(f"Average book length: {df['length'].mean():,.0f} characters")
- Downloads last month
- 23