---
license: apache-2.0
library_name: sklearn
pipeline_tag: text-classification
tags:
  - keyword-intent
  - search-intent
  - seo
---

# Keyword Intent Classifier

A search-intent classification model trained on **1,000,000 keyword records (10 lakh rows)** for SEO keyword analysis.

The model classifies keywords into four categories:

| Label | Meaning | Example |
|---|---|---|
| informational | Seeking information or answers | what is artificial intelligence |
| navigational | Looking for a specific website or page | facebook login |
| commercial | Researching or comparing before buying | best crm software |
| transactional | Intending to purchase or take action | buy laptop online |

## Installation

Use Python 3.12. Download `requirements.txt` from this repository and install the dependencies:

```bash
python -m pip install -r requirements.txt
```

## Usage

Download the model and its required supporting module, then predict keyword intent:

```python
import sys
from pathlib import Path

import joblib
from huggingface_hub import snapshot_download

folder = snapshot_download(
    repo_id="Egrove/keyword-intent-classifier",
    allow_patterns=[
        "keyword_intent_model.pkl",
        "intent_preprocess.py",
    ],
)

sys.path.insert(0, folder)

model = joblib.load(
    Path(folder) / "keyword_intent_model.pkl"
)

keywords = [
    "facebook login",
    "buy laptop online",
    "what is artificial intelligence",
    "best crm software",
]

predictions = model.predict(keywords)

for keyword, intent in zip(keywords, predictions):
    print(f"{keyword} -> {intent}")
```

### Example output

```text
facebook login -> navigational
buy laptop online -> transactional
what is artificial intelligence -> informational
best crm software -> commercial
```

Keep `intent_preprocess.py` available when loading the model.

## Limitations

Keywords can have multiple or ambiguous meanings. Evaluate predictions on your own keywords before production use. No benchmark accuracy or F1 score is claimed.

## Security

Only load this model and its supporting Python module if you trust their source. Loading pickle/joblib files and importing Python modules can execute code.

## License

This model is released under the **Apache License, Version 2.0**. See [LICENSE.txt](LICENSE.txt) for the full terms.