Cosine Similarity¶

Libraries¶

In [1]:

# 3pps
import numpy as np

# 3pps import numpy as np

Functions¶

In [2]:

def normalizar_matriz(matriz: np.ndarray) -> np.ndarray:
    return matriz/np.expand_dims((np.sqrt(np.sum(np.power(matriz, 2), axis=1))), axis=-1)


def cosine_similarity(matriz: np.ndarray) -> np.ndarray:
    return matriz @ matriz.T

def normalizar_matriz(matriz: np.ndarray) -> np.ndarray: return matriz/np.expand_dims((np.sqrt(np.sum(np.power(matriz, 2), axis=1))), axis=-1) def cosine_similarity(matriz: np.ndarray) -> np.ndarray: return matriz @ matriz.T

Main¶

In [3]:

X = np.array([
    [1, 2, 3],  
    [4, 5, 6],   
    [1, 0, 0],  
    [0, 1, 0]   
], dtype=float)

print("Embeddings originales:\n", X)

X_normalized = normalizar_matriz(matriz=X)
print("\nEmbeddings normalizados:\n", X_normalized)

similarity_matrix = cosine_similarity(matriz=X_normalized)
print("\nMatriz de similitud:\n", similarity_matrix)

X = np.array([ [1, 2, 3], [4, 5, 6], [1, 0, 0], [0, 1, 0] ], dtype=float) print("Embeddings originales:\n", X) X_normalized = normalizar_matriz(matriz=X) print("\nEmbeddings normalizados:\n", X_normalized) similarity_matrix = cosine_similarity(matriz=X_normalized) print("\nMatriz de similitud:\n", similarity_matrix)

Embeddings originales:
 [[1. 2. 3.]
 [4. 5. 6.]
 [1. 0. 0.]
 [0. 1. 0.]]

Embeddings normalizados:
 [[0.26726124 0.53452248 0.80178373]
 [0.45584231 0.56980288 0.68376346]
 [1.         0.         0.        ]
 [0.         1.         0.        ]]

Matriz de similitud:
 [[1.         0.97463185 0.26726124 0.53452248]
 [0.97463185 1.         0.45584231 0.56980288]
 [0.26726124 0.45584231 1.         0.        ]
 [0.53452248 0.56980288 0.         1.        ]]