📄 Text Classification Dataset

Multi-class text classification dataset with labeled documents for news categorization, topic classification, and document analysis.

---

📋 Project Information

Property	Value
Project	Text Classification Dataset
Category	Text Data / NLP
Author	Molla Samser
Designer & Tester	Rima Khatun
Website	https://rskworld.in
Email	help@rskworld.in
Phone	+91 93305 39277

---

📖 Description

This dataset includes labeled documents across multiple categories for text classification tasks. Perfect for:

• 📰 News Categorization - Classify news articles into categories
• 🏷️ Topic Classification - Identify main topics from text
• 📑 Document Analysis - Analyze and categorize documents
• 🤖 NLP Model Training - Train and fine-tune models

---

✨ Features

Core Features

• ✅ Multiple document categories (6 classes)
• ✅ Large labeled dataset (240+ training samples)
• ✅ Train/Validation/Test splits
• ✅ Multiple formats (CSV, JSON, TXT)
• ✅ Transformer ready format (BERT, RoBERTa)

🆕 Advanced Features

• 🔥 Interactive Data Explorer - Visual data exploration tool
• 🔥 REST API Server - Flask-based prediction API
• 🔥 Data Augmentation - 6 augmentation techniques
• 🔥 Model Explainability - LIME-based explanations
• 🔥 Batch Processing - High-throughput classification
• 🔥 Advanced Visualizations - Word clouds, confusion matrices
• 🔥 Performance Benchmarking - Model comparison tools
• 🔥 Cross-Validation - Robust model evaluation

---

📊 Dataset Statistics

Metric	Value
Training Samples	240
Validation Set	30
Test Set	30
Categories	6
Avg. Text Length	~20 words

Categories

Label	Category	Description	Color
0	Technology	Tech news, gadgets, software, AI	🔵 Blue
1	Sports	Athletics, competitions, leagues	🟢 Green
2	Politics	Government, policy, elections	🟣 Purple
3	Entertainment	Movies, music, TV shows, celebrities	🩷 Pink
4	Business	Finance, markets, economy	🟡 Amber
5	Science	Research, discoveries, space, health	🔵 Cyan

---

🛠️ Technologies

---

📁 Project Structure

text-classification/
├── index.html                  # Main showcase page
├── explorer.html               # 🆕 Interactive data explorer
├── README.md                   # Documentation
├── requirements.txt            # Python dependencies
├── text-classification.svg     # Project logo
│
├── assets/
│   ├── css/
│   │   └── style.css           # Styles
│   ├── js/
│   │   └── main.js             # Scripts
│   └── favicon.svg             # Favicon
│
├── data/
│   ├── csv/
│   │   ├── train.csv           # Training data (240 samples)
│   │   ├── validation.csv      # Validation data
│   │   ├── test.csv            # Test data
│   │   └── full_dataset.csv    # Complete dataset
│   ├── json/
│   │   ├── dataset.json        # JSON format
│   │   └── full_dataset.json   # Complete JSON
│   └── txt/
│       ├── categories.txt      # Category labels
│       └── sample_documents.txt
│
├── scripts/
│   ├── preprocessing.py        # Text preprocessing
│   ├── train_classifier.py     # Traditional ML training
│   ├── train_transformers.py   # BERT/Transformer training
│   ├── data_augmentation.py    # 🆕 6 augmentation techniques
│   ├── visualizations.py       # 🆕 Word clouds, charts
│   ├── api_server.py           # 🆕 REST API server
│   ├── model_explainability.py # 🆕 LIME explanations
│   └── batch_processor.py      # 🆕 Batch classification
│
└── notebooks/
    └── text_classification_tutorial.ipynb  # Complete tutorial

---

🚀 Quick Start

1. Clone or Download

# Download the dataset
wget https://rskworld.in/datasets/text-classification.zip
unzip text-classification.zip
cd text-classification

2. Install Dependencies

pip install -r requirements.txt

3. Load Dataset

import pandas as pd

# Load training data
train_df = pd.read_csv('data/csv/train.csv', comment='#')
print(f"Training samples: {len(train_df)}")
print(train_df.head())

4. Train a Model

# Traditional ML model
python scripts/train_classifier.py

# View visualizations
python scripts/visualizations.py ../data

---

🆕 Advanced Features Usage

📊 Interactive Data Explorer

Open explorer.html in your browser to:

• Filter documents by category
• Search through the dataset
• View category distribution charts
• Analyze word count distributions

🌐 REST API Server

# Start the API server
cd scripts
python api_server.py --demo --port 5000

API Endpoints:

GET  /                  - API info
GET  /health            - Health check
GET  /categories        - List all categories
POST /predict           - Classify single text
POST /predict/batch     - Classify multiple texts
POST /analyze           - Detailed text analysis

Example API Call:

curl -X POST http://localhost:5000/predict \
  -H "Content-Type: application/json" \
  -d '{"text": "Apple announces new iPhone with AI features"}'

🔄 Data Augmentation

from scripts.data_augmentation import TextAugmenter

augmenter = TextAugmenter(num_aug=5, random_state=42)
text = "Apple announces revolutionary new iPhone"

augmented = augmenter.augment(text)
for i, aug_text in enumerate(augmented, 1):
    print(f"{i}. {aug_text}")

Supported Techniques:

• Synonym Replacement (SR)
• Random Insertion (RI)
• Random Swap (RS)
• Random Deletion (RD)
• Character-level augmentation
• Keyboard error simulation

🔍 Model Explainability

from scripts.model_explainability import TextExplainer

explainer = TextExplainer(classifier_fn)
explanation = explainer.explain("New AI-powered smartphone released")

print(f"Predicted: {explanation['predicted_category']}")
print("Important words:")
for item in explanation['word_importance'][:5]:
    print(f"  {item['word']}: {item['importance']:.4f}")

📦 Batch Processing

# Process a file of texts
python scripts/batch_processor.py process \
  --input input.csv \
  --output predictions.csv \
  --model model.joblib \
  --batch-size 100

# Evaluate predictions
python scripts/batch_processor.py evaluate \
  --predictions predictions.csv \
  --ground-truth ground_truth.csv \
  --output report.json

📈 Visualizations

# Generate all visualizations
python scripts/visualizations.py ../data

# Outputs:
# - visualizations/category_distribution.png
# - visualizations/text_length_distribution.png
# - visualizations/wordcloud_all.png
# - visualizations/wordclouds_by_category/

---

📝 Usage Examples

Basic Classification

from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model import LogisticRegression
import pandas as pd

# Load data
train_df = pd.read_csv('data/csv/train.csv', comment='#')

# Vectorize
tfidf = TfidfVectorizer(max_features=10000)
X = tfidf.fit_transform(train_df['text'])
y = train_df['label']

# Train
model = LogisticRegression()
model.fit(X, y)

# Predict
text = "Apple unveils new iPhone with AI features"
prediction = model.predict(tfidf.transform([text]))
print(f"Predicted: {prediction[0]}")  # 0 = Technology

Using Transformers (BERT)

from transformers import AutoTokenizer, AutoModelForSequenceClassification

# Load model
model_name = "bert-base-uncased"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name, num_labels=6)

# Tokenize
text = "Scientists discover new planet in nearby galaxy"
inputs = tokenizer(text, return_tensors="pt", padding=True, truncation=True)

# Predict
outputs = model(**inputs)
prediction = outputs.logits.argmax().item()
print(f"Predicted label: {prediction}")  # 5 = Science

---

📊 Model Performance

Model	Accuracy	F1 Score	Inference (ms)
Naive Bayes	85.2%	0.847	~1ms
Logistic Regression	89.7%	0.892	~2ms
Linear SVM	88.9%	0.885	~2ms
BERT (fine-tuned)	94.3%	0.941	~50ms

---

📜 License

This dataset is provided for educational purposes only.

Copyright (c) 2026 RSK World - All Rights Reserved

---

👨‍💻 Author

Molla Samser

• 🌐 Website: https://rskworld.in
• 📧 Email: help@rskworld.in
• 📱 Phone: +91 93305 39277

Designer & Tester

Rima Khatun

---

🤝 Support

If you have any questions or need support:

• 📧 Email: support@rskworld.in
• 🌐 Contact: https://rskworld.in/contact.php

---

🔗 Links

• Homepage
• Data Explorer
• Download Dataset

---

Made with ❤️ by RSK World
rskworld.in