MODEL_CARD.md

# Boolean Search Query Model

This model is fine-tuned to convert natural language queries into boolean search expressions, optimized for academic and research database searching.

## Model Details

- **Base Model**: Meta-Llama-3.1-8B
- **Training Type**: LoRA fine-tuning
- **Task**: Converting natural language to boolean search queries
- **Languages**: English
- **License**: Same as base model

## Intended Use

- Converting natural language search requests into proper boolean expressions
- Academic and research database searching
- Information retrieval query formulation

## Performance

### Test Results

Base Model vs Fine-tuned Model comparison:

```
Natural Query: "Studies examining the relationship between exercise and mental health"
Base: exercise AND mental health
Fine-tuned: exercise AND "mental health"  # Properly handles multi-word terms

Natural Query: "Articles about artificial intelligence ethics and regulation or policy"
Base: "artificial intelligence ethics" AND ("regulation" OR "policy")  # Treats AI ethics as one concept
Fine-tuned: "artificial intelligence" AND (ethics OR regulation OR policy)  # Properly splits concepts
```

### Key Improvements

1. Meta-term Removal
   - Automatically removes terms like "articles", "papers", "research", "studies"
   - Focuses on actual search concepts

2. Proper Term Quoting
   - Only quotes multi-word phrases
   - Single words remain unquoted

3. Logical Grouping
   - Appropriate use of parentheses for OR groups
   - Clear operator precedence

4. Minimal Formatting
   - No unnecessary parentheses
   - No duplicate terms

## Limitations

- English language only
- May not handle specialized domain terminology optimally
- Limited to boolean operators (AND, OR, NOT)
- Designed for academic/research context

## Training Data

The model was trained on a curated dataset of natural language queries paired with their correct boolean translations. Dataset characteristics:

- Size: 135 examples
- Format: Natural query → Boolean expression pairs
- Source: Manually curated academic search examples
- Validation: Expert-reviewed for accuracy

## Training Process

- **Method**: LoRA fine-tuning
- **Hardware**: NVIDIA GeForce RTX 4070 Ti SUPER

## How to Use

```python
from unsloth import FastLanguageModel

# Load model
model, tokenizer = FastLanguageModel.from_pretrained(
    "Zwounds/boolean-search-model",
    max_seq_length=2048,
    dtype=None,  # Auto-detect
    load_in_4bit=True
)
FastLanguageModel.for_inference(model)

# Format query
query = "Find papers about climate change and renewable energy"
formatted = f"""Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.

### Instruction:
Convert this natural language query into a boolean search query by following these rules:

1. FIRST: Remove all meta-terms from this list (they should NEVER appear in output):
   - articles, papers, research, studies
   - examining, investigating, analyzing
   - findings, documents, literature
   - publications, journals, reviews
   Example: "Research examining X" → just "X"

2. SECOND: Remove generic implied terms that don't add search value:
   - Remove words like "practices," "techniques," "methods," "approaches," "strategies"
   - Remove words like "impacts," "effects," "influences," "role," "applications"
   - For example: "sustainable agriculture practices" → "sustainable agriculture"
   - For example: "teaching methodologies" → "teaching"
   - For example: "leadership styles" → "leadership"

3. THEN: Format the remaining terms:
   CRITICAL QUOTING RULES:
   - Multi-word phrases MUST ALWAYS be in quotes - NO EXCEPTIONS
   - Examples of correct quoting:
     - Wrong: machine learning AND deep learning
     - Right: "machine learning" AND "deep learning"
     - Wrong: natural language processing
     - Right: "natural language processing"
   - Single words must NEVER have quotes (e.g., science, research, learning)
   - Use AND to connect required concepts
   - Use OR with parentheses for alternatives (e.g., ("soil health" OR biodiversity))

Example conversions showing proper quoting:
"Research on machine learning for natural language processing"
→ "machine learning" AND "natural language processing"

"Studies examining anxiety depression stress in workplace"
→ (anxiety OR depression OR stress) AND workplace

"Articles about deep learning impact on computer vision"
→ "deep learning" AND "computer vision"

"Research on sustainable agriculture practices and their impact on soil health or biodiversity"
→ "sustainable agriculture" AND ("soil health" OR biodiversity)

"Articles about effective teaching methods for second language acquisition"
→ teaching AND "second language acquisition"

### Input:
{query}

### Response:
"""

# Generate boolean query
inputs = tokenizer(formatted, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=100)
result = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(result)  # "climate change" AND "renewable energy"
```

## Citation

If you use this model in your research, please cite:
```bibtex
@misc{boolean-search-llm,
  title={Boolean Search Query LLM},
  author={Stephen Zweibel},
  year={2025},
  publisher={Hugging Face},
  url={https://huggingface.co/Zwounds/boolean-search-model}
}