Extract the top single words or 2- and 3-word phrases from any text, ranked by frequency and share, with adjustable stopword and length filters.
"Keyword extraction" means two different jobs depending on who is asking. An SEO writer wants the single words a page repeats too often. A researcher combing through interview notes wants the two- and three-word phrases that keep recurring: "customer support," not "customer" and "support" scored apart. The Keyword & Phrase Extractor switches between both with one control, Phrase Length.
Paste a draft into Paste Your Content and results rank as you type, filtered by two thresholds you can tune yourself: how short a word can be before it's ignored, and how many times a term has to repeat before it counts as a keyword at all. Common words like "the" and "of" are stripped by default, and for phrases they are only stripped from the edges. A middle preposition survives inside "quality of life," but "the" still falls off the front of "the quality of."
In 1-word mode, a term's share is its count divided by the total word count of the passage: the lead sample's 81-word fox/dog passage puts "dog" and "fox" tied at 4× each, 4.9% apiece, with "over," "quick" and "wild" trailing at 2× and 2.5%. In 2- and 3-word mode the denominator changes to the number of overlapping windows, total words minus phrase length plus one, not the raw word count, so a phrase's percentage is not directly comparable to a single word's from the same passage. A phrase is kept only when neither its first nor its last word is a stopword, which is why "of the fox" can never rank but "brown fox jumps" can.
480/20,000 characters · 81 words analyzed
Phrase length