Paste an article, product description or notes to pull out the words and phrases that matter most. Choose RAKE for multi-word key phrases or TF-IDF for distinctive terms, and copy the results as a list or CSV.
Last updated:
Advertisement
Advertisement
How to use the Keyword Extractor
Paste your text: an article, page copy, transcript or a set of reviews. Longer texts give better results.
Pick a method: RAKE for key phrases, TF-IDF for terms that stand out in parts of the text, or Frequency for simple counts.
Choose what to show: single words, phrases or both, and how many results.
Refine with custom stop words, minimum word length and plural grouping.
Copy or export the keywords as a list, comma-separated tags or a CSV file.
How the three methods score keywords
Method
Formula
Best for
RAKE
word score = degree ÷ frequency; phrase score = sum of its word scores, boosted by 1 + ln(occurrences)
Descriptive multi-word key phrases, tags, titles
TF-IDF
(count ÷ total words) × (ln((N + 1) ÷ (df + 1)) + 1), with a bonus of 50% per extra word in a phrase
Distinctive topics in long texts with several sections
Frequency
count, with a 25% bonus per extra word in a phrase
Quick view of the most-used content words
In RAKE, the frequency of a word is how many candidate phrases it appears in and its degree is the total length of those phrases, so a word that mostly appears inside longer phrases scores higher. In TF-IDF, N is the number of documents (paragraphs, or sentences when there are fewer than three paragraphs) and df is how many of them contain the term. Scores in the table are shown relative to the top result (100).
Worked example
Take the sentence “Home solar panels lower electricity bills; solar panels need little maintenance.” Stop words, general verbs (lower, need) and punctuation split it into the candidate phrases home solar panels, electricity bills, solar panels and little maintenance. “solar” appears in 2 phrases whose lengths add up to 3 + 2 = 5, so its score is 5 ÷ 2 = 2.5; “panels” also scores 2.5 and “home” scores 3 ÷ 1 = 3. The phrase home solar panels therefore scores 3 + 2.5 + 2.5 = 8, ahead of solar panels at 5 (before the small boost for repeated phrases).
Tips for better keywords
Use at least a few paragraphs. Statistical scores are unreliable on a single short sentence.
Add domain stop words such as your brand or words like “page” and “click” that appear everywhere but say nothing about the topic.
Raise the minimum word length to 4 when short words like “use” or “set” clutter the list.
Compare methods. Keywords that rank high in both RAKE and TF-IDF are usually the strongest candidates for tags and meta keywords.
Frequently asked questions
How does the keyword extractor work?
It splits your text into words, removes stop words such as “the”, “and” and “with”, and scores what is left. RAKE scores phrases by how often their words appear and how many other words they co-occur with; TF-IDF scores terms that are frequent overall but concentrated in some paragraphs; Frequency simply counts.
What is RAKE?
RAKE (Rapid Automatic Keyword Extraction) treats the runs of words between stop words and punctuation as candidate phrases. Each word gets a score of degree ÷ frequency, where degree counts the words it appears alongside, and a phrase's score is the sum of its word scores. Longer, specific phrases tend to rank high.
What is TF-IDF?
TF-IDF multiplies term frequency (how often a term appears) by inverse document frequency (how rare it is across documents). Here each paragraph, or each sentence if there are fewer than three paragraphs, counts as a document, so terms that are frequent but appear everywhere score lower than terms that define particular sections.
Which method should I use?
Use RAKE to find descriptive key phrases for tags, titles or summaries. Use TF-IDF to find the distinctive topics in a long document with several sections. Use Frequency when you just want the most-used content words.
How is this different from a keyword density checker?
A density checker counts how often given words appear so you can check SEO balance. A keyword extractor ranks which words and phrases best describe the text, even ones that appear only a few times.
Does it work for languages other than English?
The word splitting works for any language that uses spaces, but the built-in stop word list is English. For other languages add common words to the custom stop word box for cleaner results.
Are all the tools on FreeToolCabin tested for quality and accuracy?
Yes! Every tool on our site undergoes automated Q/A testing in real browser environments to ensure high precision, correct output, and zero errors. If you happen to spot a bug or have a suggestion, please contact us at [email protected] and our team will review it right away.