# README: Bed Stories Dataset

## Description
The Bed Stories dataset is part of a research project that focuses on improving children's information retrieval through semantic query expansion using WordNet. It contains a curated collection of 200 English children’s stories and a set of sample child-formulated queries. This dataset supports the implementation and evaluation of a Lucene-based search engine tailored to children's information-seeking behavior.

## Size
Total dataset size: 0.28 MB

## Platform
The dataset is intended for use with:
- Lucene-based search engine platforms
- Python-based tools (optional for preprocessing and evaluation)

## Environment
To use this dataset effectively, the following environment is recommended:
- Operating System: Windows 10+, macOS, or Linux
- Java JDK 8 or later (for Lucene)
- Python 3.7+ (optional, for data preprocessing)
- Libraries: NLTK (with WordNet), Pandas (for evaluation), Apache Lucene

## Major Component Description
- `Data/Queries.txt`: A text file containing a list of child-generated queries. Each line represents one query.
- `Data/Data Collection/`: A folder with 200 plain-text story files (e.g., `Text 1.txt`, `Text 2.txt`, ...). Each file contains one English children's story relevant to common themes like honesty, family, friendship, etc.

## Detailed Setup Instructions
1. **Unzip** the dataset archive (`Dataset.zip`) to your desired working directory.
2. **Directory Structure:**
   ```
   Dataset/
   └── Data/
       ├── Queries.txt
       └── Data Collection/
           ├── Text 1.txt
           ├── Text 2.txt
           └── ... (200 total story files)
   ```
3. **Prepare Lucene Index**:
   - Set up an indexing script using Apache Lucene.
   - Point it to the `Data Collection/` directory to index all `.txt` files.

4. **Query Execution**:
   - Read queries from `Queries.txt`.
   - Expand each query using WordNet (with NLTK in Python or integrated in Java).
   - Use the expanded queries for search via Lucene’s API.

5. (Optional) **Evaluation**:
   - Compare retrieved results against predefined relevance judgments using MAP or P@3 metrics.

## Contact Information
For questions or collaboration, please contact:
- Wasmiah Alyami: wwhhww22@outlook.sa
- Hend Alzahrani: hend1411@gmail.com