---
license: mit
task_categories:
- text-classification
- text-generation
- summarization
- question-answering
- feature-extraction
language:
- en
tags:
- technology-news
- techcrunch
- news-dataset
- journalism
- media-analysis
- bias-analysis
- content-generation
- nlp
- business-intelligence
- startup-news
- funding-analysis
size_categories:
- 10K<n<100K
source_datasets:
- original
multilinguality:
- monolingual
pretty_name: TechCrunch News Articles Dataset
dataset_info:
  features:
  - name: title
    dtype: string
  - name: content
    dtype: string
  - name: publish_date
    dtype: timestamp[s]
  - name: url
    dtype: string
  - name: word_count
    dtype: int64
  - name: quality_score
    dtype: float64
  - name: content_hash
    dtype: string
  - name: bias_analysis
    dtype: string
  num_rows: 10265
  num_examples: 10265
  download_size: 54294941
  dataset_size: 27560611
configs:
- config_name: default
  data_files:
  - split: train
    path: "data/techcrunch.parquet"
viewer: true
---

# TechCrunch News Articles Dataset

## 📊 Dataset Overview

This dataset contains **10,265 high-quality news articles** scraped from TechCrunch, one of the leading technology news websites. The dataset includes comprehensive article content, metadata, and quality assessments suitable for various NLP tasks including text classification, sentiment analysis, summarization, and content generation.

## 🎯 Key Features

- **10,265 articles** with full text content
- **High-quality filtering** with quality scores (avg: 0.96)
- **Content cleaned** and optimized for ML applications
- **Rich metadata** including publish date, bias analysis, and quality metrics
- **Comprehensive bias analysis** for ethical AI development
- **Ready for NLP tasks** including classification, summarization, and generation
- **HuggingFace optimized** format for seamless integration

## 📈 Dataset Statistics

| Metric | Value |
|--------|-------|
| Total Articles | 10,265 |
| Average Content Length | 4,263 characters |
| Average Word Count | 686 words |
| Average Quality Score | 0.957 |
| Date Range | 2023-04-10 to 2025-07-09 |
| File Size (JSON) | 51.8 MB |
| File Size (Parquet) | 26.3 MB |

## 🏗️ Data Structure

The dataset contains the following columns:

- **`title`**: Article headline
- **`content`**: Full article text content (cleaned)
- **`publish_date`**: Publication timestamp
- **`url`**: Original article URL
- **`word_count`**: Number of words in content
- **`quality_score`**: Automated quality assessment (0-1)
- **`content_hash`**: MD5 hash for deduplication
- **`bias_analysis`**: JSON string containing comprehensive bias assessment

## 🔍 Quality Assurance

This dataset implements multiple quality control measures:

- **Content Filtering**: Minimum 16 words per article
- **Content Cleaning**: Navigation elements and website artifacts removed
- **Quality Scoring**: Automated assessment based on content length, structure, and readability
- **Deduplication**: Content hash-based duplicate removal
- **Bias Analysis**: Comprehensive bias detection and scoring
- **Manual Review**: Sample validation for quality assurance

### Quality Score Distribution

- **Minimum**: 0.600
- **Average**: 0.957  
- **Maximum**: 1.000

## 🚀 Getting Started

### Loading the Dataset

```python
# Using datasets library (recommended)
from datasets import load_dataset
dataset = load_dataset('abhilash88/techcrunch-articles')

# Using pandas for Parquet format
import pandas as pd
df = pd.read_parquet('path/to/techcrunch_articles.parquet')

# Access the training split
df = dataset['train'].to_pandas()
```

### Basic Analysis

```python
# Dataset overview
print(f"Total articles: {len(dataset['train']):,}")
print(f"Features: {dataset['train'].features}")

# Convert to pandas for analysis
df = dataset['train'].to_pandas()
print(f"Average word count: {df['word_count'].mean():.0f}")
print(f"Date range: {df['publish_date'].min()} to {df['publish_date'].max()}")

# Quality distribution
print(f"Quality score distribution:")
print(df['quality_score'].describe())
```

## 📜 Legal and Attribution

### Data Source
- **Original Source**: TechCrunch (techcrunch.com)
- **Scraping Method**: Respectful automated collection with rate limiting
- **Usage Rights**: Educational and research purposes
- **Attribution**: Please cite this dataset in academic work

### Recommended Citation

```bibtex
@dataset{techcrunch_articles_2025,
  title={TechCrunch News Articles Dataset},
  author={Abhilash Sahoo},
  year={2025},
  publisher={Hugging Face},
  url={https://huggingface.co/datasets/abhilash88/techcrunch-articles},
  note={10,265 high-quality technology news articles with bias analysis}
}
```

## 📞 Contact

- **Creator**: Abhilash Sahoo
- **Hugging Face**: [@abhilash88](https://huggingface.co/abhilash88)
- **Issues**: Please use the dataset discussion section

---

**🚀 Ready to build amazing NLP applications with high-quality technology news content!**