TF-IDF

Gold definitionUpdated Apr 2, 2026

Definition

TF-IDF (Term Frequency-Inverse Document Frequency) is a numerical statistic reflecting how important a word is to a document in a collection or corpus. It's widely used in information retrieval and text mining to weigh terms based on their relevance.

At a glance

Executive summary

TF-IDF is a method to quantify how important a word is to a document within a larger collection of documents. It helps computers understand which words are key to a document's meaning, making it useful for tasks like finding relevant information or classifying text.

TL;DR

TF-IDF assigns a score to words in documents to show how important they are, helping computers understand text better for tasks like searching or categorizing.

Key points

Calculates a numerical weight for each term by multiplying its frequency in a document (TF) by its inverse frequency across the corpus (IDF).
Quantifies term importance in documents, enabling effective text representation for information retrieval, classification, and filtering.
Used by NLP researchers, data scientists, and machine learning engineers in text analysis, information retrieval, and document classification.
Simpler than neural embeddings (like Word2Vec or BERT) as it's count-based and doesn't capture semantic relationships, but computationally lighter and often effective for specific tasks.
While foundational, it's often combined with or superseded by deep learning embeddings for complex semantic tasks, but remains a strong baseline and feature engineering tool.

Use cases

Ranking web pages by relevance to a user's query in search engines.
Identifying suspicious emails based on characteristic word patterns for spam or phishing detection.
Categorizing news articles into topics (e.g., sports, politics) or detecting PII leakage using classifiers trained on TF-IDF features.
Automatically identifying the most important keywords from a document for summarization or tagging.
Comparing documents by their TF-IDF vectors to find similarities in content for plagiarism detection.

Also known as

TFIDF, Term Frequency-Inverse Document Frequency, Term Weighting