Cosine Similarity Calculator

Understanding Cosine Similarity

Cosine similarity is a metric used to measure how similar two vectors are, regardless of their magnitude. It calculates the cosine of the angle between two vectors in a multi-dimensional space. The result ranges from -1 (exactly opposite) through 0 (orthogonal/uncorrelated) to 1 (exactly the same direction).

Unlike Euclidean distance, cosine similarity focuses on the orientation of vectors rather than their magnitude. This makes it especially valuable when the length of the vector is less important than its direction, which is common in text analysis and information retrieval.

The Cosine Similarity Formula

Cosine Similarity

The ratio of the dot product to the product of magnitudes.

cos(θ) = (A · B) / (||A|| · ||B||)

Dot Product

Sum of element-wise products of the two vectors.

A · B = Σ a_i × b_i

Vector Magnitude

The Euclidean norm (L2 norm) of the vector.

||A|| = sqrt(Σ a_i²)

Cosine Distance

A dissimilarity measure derived from cosine similarity.

distance = 1 - similarity

Interpreting Cosine Similarity Values

1.0: The vectors point in exactly the same direction. They are identical in orientation (perfectly similar).
0.0: The vectors are orthogonal (perpendicular). There is no linear correlation between them.
-1.0: The vectors point in exactly opposite directions. They are diametrically dissimilar.
0.5 to 1.0: High similarity. Common threshold for "similar" in many applications.
0.0 to 0.5: Low to moderate similarity.

Applications of Cosine Similarity

Natural Language Processing (NLP): Cosine similarity is the standard metric for comparing document vectors, word embeddings (Word2Vec, GloVe), and sentence embeddings. Two documents with similar topics will have high cosine similarity in their TF-IDF or bag-of-words representations.
Recommendation Systems: Used to find similar users or items in collaborative filtering. Netflix, Spotify, and Amazon all use variants of cosine similarity in their recommendation engines.
Information Retrieval: Search engines use cosine similarity to rank documents by relevance to a query vector.
Machine Learning: Used as a loss function component in contrastive learning, and as a similarity metric in k-nearest neighbors and clustering algorithms.
Image Recognition: Feature vectors from convolutional neural networks are compared using cosine similarity for face recognition and image search.
Plagiarism Detection: Documents are converted to vectors and compared using cosine similarity to detect copied content.

Cosine Similarity vs. Euclidean Distance

While Euclidean distance measures the straight-line distance between two points in space, cosine similarity measures the angle between two vectors. This distinction matters significantly: two documents about the same topic but of different lengths will be far apart in Euclidean distance but close in cosine similarity. This is why cosine similarity is preferred for high-dimensional sparse data like text.

Cosine Similarity in Python

In practice, cosine similarity is implemented using libraries like NumPy and scikit-learn. The formula can be computed efficiently using vectorized operations: np.dot(A, B) / (np.linalg.norm(A) * np.linalg.norm(B)). Scikit-learn provides cosine_similarity() for batch computation.

Tips for Using Cosine Similarity

Both vectors must have the same number of dimensions.
Zero vectors (all components zero) will cause division by zero. Handle this edge case in your code.
For non-negative vectors (like TF-IDF), cosine similarity ranges from 0 to 1.
Cosine similarity is invariant to scaling: multiplying a vector by a positive constant does not change its cosine similarity with other vectors.
For normalized vectors (unit vectors), cosine similarity equals the dot product.

Enter Vector Components

Result

Step-by-Step Solution