---
language:
- en
license: mit
size_categories:
- n<1K
task_categories:
- text-generation
tags:
- geo
- seo
- search-engine-optimization
- query-generation
---

# GEO Agent Dataset with Queries

Dataset with generated train/test queries for Generative Engine Optimization (GEO) research.

## Dataset Description

This dataset contains 208 web documents with automatically generated search queries for training and evaluation.

## Features

| Column | Description |
|--------|-------------|
| doc_id | Unique document identifier |
| url | Source URL |
| raw_html | Original HTML content |
| cleaned_text | Parsed plain text content |
| cleaned_text_length | Character count of cleaned text |
| tags | Topic classification tags |
| primary_topic | Main topic category |
| data_source | Original data source |
| query | Original search query |
| train_queries | Generated training queries (~20 per doc) |
| test_queries | Generated test queries (~36 per doc) |

## Statistics

- **Total documents**: 208
- **Avg train queries**: 19.7 per document
- **Avg test queries**: 35.9 per document
- **Total train queries**: 4096
- **Total test queries**: 7457

## Usage

```python
from datasets import load_dataset

ds = load_dataset("erv1n/GEO_Agent_with_queries")

# Access queries
for example in ds["train"]:
    print(f"Doc: {example['doc_id']}")
    print(f"Train queries: {example['train_queries'][:3]}")
    print(f"Test queries: {example['test_queries'][:3]}")
    break
```

## Related Datasets

- [erv1n/GEO_Agent](https://huggingface.co/datasets/erv1n/GEO_Agent) - Base dataset without queries