auto-keyword-ranker

A lightweight Python library to automatically extract and rank the most relevant keywords and keyphrases from text.

Goal: Provide a single-line call to obtain ranked keywords for articles, blog posts, or short documents. Uses TF-IDF by default, with optional re-ranking via sentence-transformer embeddings.

Installation

pip install auto-keyword-ranker

With optional embedding-based re-ranking:

pip install auto-keyword-ranker[embed]

Quickstart

from autokeyword import rank_keywords

text = """
Artificial intelligence is transforming industries by enabling new capabilities
such as natural language processing, computer vision, and advanced data analytics.
Companies across healthcare, finance, and manufacturing are investing heavily in AI
to automate decision-making, enhance efficiency, and unlock new insights from
large datasets.
"""

# Simple TF-IDF keyword ranking
keywords = rank_keywords(text, top_n=10)
print(keywords)

Output

A list of (keyword, score) pairs, for example:

[('new', 0.25),
 ('vision advanced', 0.12),
 ('unlock new', 0.12),
 ('unlock', 0.12),
 ('transforming industries', 0.12),
 ('vision', 0.12),
 ('transforming', 0.12),
 ('processing computer', 0.12),
 ('new insights', 0.12),
 ('processing', 0.12)]

API

rank_keywords(
    texts,
    top_n=10,
    method='tfidf',
    ngram_range=(1,2),
    stop_words=True,
    use_embeddings=False,
    embedding_model=None,
    combine_score_alpha=0.6
)

See docstrings in core.py for full parameter descriptions.

CLI

You can also run the CLI (after installation):

python -m autokeyword.cli --text "Your article text here" --top 10

How It Works

TF-IDF scoring formula:

$$ \text{TF-IDF}(t, d) = \text{TF}(t, d) \times \log \frac{N}{1 + \text{DF}(t)} $$

Where:

• ( t ) = term
• ( d ) = document
• ( N ) = total number of documents

Use Cases

• SEO keyword extraction for blog posts
• Automatic tagging in content management systems
• Quick summarization of research papers
• Preprocessing for search or recommendation engines

License

MIT License © 2025 Reya Oberoi