Document similarity ranking in NLP - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to compute cosine similarity between two vectors.

NLP

from sklearn.metrics.pairwise import [1]

vec1 = [[1, 2, 3]]
vec2 = [[4, 5, 6]]
similarity = [1](vec1, vec2)
print(similarity)

Drag options to blanks, or click blank then click option'

Acosine_similarity

Beuclidean_distance

Cmanhattan_distance

Ddot_product

Attempts:

3 left

💡 Hint

Common Mistakes

Using distance functions like euclidean_distance instead of similarity.

Trying to use dot_product which is not a sklearn function.

The cosine_similarity function from sklearn.metrics.pairwise calculates the cosine similarity between vectors, which is commonly used to measure document similarity.

2fill in blank

medium

Complete the code to convert text documents into TF-IDF vectors.

NLP

from sklearn.feature_extraction.text import [1]

corpus = ['I love machine learning', 'Machine learning is fun']
tfidf_vectorizer = [1]()
tfidf_matrix = tfidf_vectorizer.fit_transform(corpus)
print(tfidf_matrix.toarray())

Drag options to blanks, or click blank then click option'

AHashingVectorizer

BTfidfVectorizer

CCountVectorizer

DDictVectorizer

Attempts:

3 left

💡 Hint

Common Mistakes

Using CountVectorizer which only counts word occurrences.

Using HashingVectorizer which does not compute TF-IDF.

TfidfVectorizer converts text documents into TF-IDF feature vectors, which reflect the importance of words in documents.

3fill in blank

hard

Fix the error in the code to correctly compute similarity scores between documents.

NLP

from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import [1]

texts = ['Data science is cool', 'I love data science']
vectorizer = TfidfVectorizer()
matrix = vectorizer.fit_transform(texts)
scores = [1](matrix)
print(scores)

Drag options to blanks, or click blank then click option'

Amanhattan_distances

Beuclidean_distances

Cpairwise_distances

Dcosine_similarity

Attempts:

3 left

💡 Hint

Common Mistakes

Using distance functions which give dissimilarity scores.

Passing the wrong matrix shape to the function.

To get similarity scores, cosine_similarity is the correct function. Distance functions return dissimilarity, not similarity.

4fill in blank

hard

Fill both blanks to create a dictionary of document similarity scores above a threshold.

NLP

from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import cosine_similarity

texts = ['AI is the future', 'AI and ML are related', 'I enjoy sports']
vectorizer = TfidfVectorizer()
matrix = vectorizer.fit_transform(texts)
sim_matrix = cosine_similarity(matrix)

threshold = 0.5
similar_docs = {i: [j for j in range(len(texts)) if sim_matrix[i][j] [1] threshold and i != j] for i in range(len(texts)) if any(sim_matrix[i][j] [2] threshold for j in range(len(texts)))}
print(similar_docs)

Drag options to blanks, or click blank then click option'

A>

B>=

C<

D<=

Attempts:

3 left

💡 Hint

Common Mistakes

Using '<' or '<=' which would select less similar documents.

Mixing different operators in the two blanks.

The code filters pairs with similarity scores strictly greater than the threshold, so both blanks use the '>' operator.

5fill in blank

hard

Fill all three blanks to rank documents by similarity to a query document.

NLP

from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import cosine_similarity

corpus = ['Deep learning is powerful', 'I like deep learning', 'Cats are cute']
vectorizer = TfidfVectorizer()
matrix = vectorizer.fit_transform(corpus)

query = ['I love learning']
query_vec = vectorizer.transform(query)
sim_scores = cosine_similarity(query_vec, matrix)[0]

ranked_docs = sorted(((i, sim_scores[i]) for i in range(len(corpus))), key=lambda x: x[1] x[2], reverse=[3])
print(ranked_docs)

Drag options to blanks, or click blank then click option'

A*

B-

C>

DTrue

Attempts:

3 left

💡 Hint

Common Mistakes

Using '*' or '+' in the key which does not sort properly.

Setting reverse=False which sorts ascending.

The sorting key uses -x[1] to sort by descending similarity scores. The reverse=True argument ensures highest scores come first.