auto-keyword-ranker 0.1.5

Lightweight keyword & keyphrase extraction and ranking using TF-IDF with optional embedding re-ranking.

auto-keyword-ranker

Lightweight Python package to extract and rank the most relevant keywords and keyphrases from text.

Goal: One-line call to get ranked keywords for articles, blog posts, or short documents.
Core approach uses TF-IDF; optional re-ranking with 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.
"""

# Simple TF-IDF keyword ranking
keywords = rank_keywords(text, top_n=5)
print(keywords)

Output

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

[('artificial intelligence', 0.42),
 ('data analytics', 0.33),
 ('natural language processing', 0.29),
 ('computer vision', 0.25),
 ('industries', 0.21)]

---

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 autokeyword/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:

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

Where:

TF(t, d) – term frequency of term t in document d

DF(t) – number of documents containing term t

N – total number of documents

---

License

MIT License © 2025 Reya Oberoi