Cosine Geometry in AI & Artificial Intelligence

When we treat items as Vectors (lists of ratings), we can visualize them in space. Euclidean Distance measures the 'Straight-line' distance between two points. If one user rates everything 5/5 and another rates everything 3/5, they will be far apart in Euclidean space. However, Cosine Similarity measures the Angle between the vectors. If both users loved Item A twice as much as Item B, their vectors point in the same direction, resulting in a high similarity score. This makes Cosine the superior choice for handling the inherent subjectivity of human ratings.

The numerator of the Cosine formula is the Dot Product. It multiplies the ratings of corresponding items and sums them up. If two items are often rated highly by the same users, the dot product will be large. We then Normalize this by dividing by the magnitudes of the vectors. This step ensures that a popular item with thousands of ratings doesn't automatically dominate the results simply because it has 'more numbers'. It scales everything to a consistent range from 0 to 1.

A common problem in RecSys is the 'Optimistic User' who gives everything 4 stars, and the 'Pessimist' who gives everything 2 stars. To the AI, the Optimist's 3 might be a 'dislike', while the Pessimist's 3 might be a 'rave review'. We solve this with Mean Centering. We subtract the user's average rating from every individual rating. Now, a positive number means 'Above Average' and a negative number means 'Below Average'. This 'Adjusted Cosine Similarity' is the industry standard for high-accuracy collaborative filtering.

How do we measure the distance between two tastes? Cosine Similarity is the gold standard for measuring how 'close' two items or users are in a multi-dimensional rating space.

Cosine similarity measures the angle between two vectors. If the angle is 0, the items are identical. If it's 90 degrees, they are unrelated.

Unlike Euclidean distance, Cosine similarity doesn't care if one user rates everything high and another rates everything low. It only cares about the *Relative* preference.

Checkpoint: What is a 'Perfect Match' score in Cosine Similarity?

• →0
• →1

We often use 'Adjusted Cosine' or 'Mean Centering' to remove user bias, ensuring that a '4' from a harsh critic means more than a '4' from a happy rater.

By mastering Cosine Similarity, you build engines that understand the nuanced patterns of human behavior, finding deep connections in the data.

Checkpoint: Why is Cosine Similarity better for ratings than Euclidean distance?

• →The math is faster
• →It ignores the 'Magnitude' (how high or low the raw numbers are) and focuses on the 'Direction' (the relative pattern of the ratings)

Cosine similarity mastered! You've learned the geometry. Ready to compress the data with Matrix Factorization and SVD?

Compute Real Cosine Similarity. Finish implementing cosine similarity and confirm two perfectly aligned vectors score 1.0.