A threat intelligence analyst explains that their ML model uses 'cosine similarity' to find alerts similar to known indicators. A junior analyst asks what cosine similarity measures. What is the best explanation?

• A.Cosine similarity measures the angle between two vectors in the embedding space, capturing semantic similarity independent of document length - two alerts with the same threat pattern expressed in different verbosity levels will have high cosine similarity because they point in the same direction in embedding space
• B.Cosine similarity counts the number of common words between two text samples, dividing by total word count
• C.Cosine similarity measures the Euclidean distance between two data points and returns a value between 0 and 100
• D.Cosine similarity is a hash comparison function that detects exact duplicate text