Evaluating generated text (BLEU, ROUGE) in NLP - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to import the BLEU score function from nltk.

NLP

from nltk.translate.bleu_score import [1]

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

Asentence_bleu

Bbleu_score

Ccorpus_bleu

Dbleu

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'bleu_score' which is the module, not the function.

Using 'corpus_bleu' which is for multiple sentences.

The function sentence_bleu is used to calculate BLEU score for a single sentence.

2fill in blank

medium

Complete the code to calculate the BLEU score for a candidate sentence.

NLP

score = sentence_bleu([reference], [1])

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

Acandidate_tokens

Breferences

Ccandidate_sentence

Dcandidate

Attempts:

3 left

💡 Hint

Common Mistakes

Passing the reference list again instead of candidate.

Using a variable name that does not exist.

The candidate sentence tokens should be passed as the second argument to sentence_bleu.

3fill in blank

hard

Fix the error in the ROUGE score calculation by completing the missing import.

NLP

from rouge_score import [1]

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

ARougeScorer

Brouge_scorer

Cscore_rouge

Drouge

Attempts:

3 left

💡 Hint

Common Mistakes

Using lowercase 'rouge_scorer' which is a module, not the class.

Using 'score_rouge' which does not exist.

The class RougeScorer is imported from the rouge_score package to calculate ROUGE scores.

4fill in blank

hard

Fill both blanks to create a ROUGE scorer and calculate scores for two texts.

NLP

scorer = [1](['rouge1', 'rougeL'])
scores = scorer.score([2], generated)

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

ARougeScorer

Breference_text

Creference

Dscore

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'reference_text' which is not defined.

Using 'score' which is a variable, not a function/class.

RougeScorer is used to create the scorer object. The reference text variable is named reference.

5fill in blank

hard

Fill all three blanks to compute BLEU and ROUGE scores and print them.

NLP

bleu = sentence_bleu([reference], [1])
scorer = [2](['rouge1', 'rougeL'])
rouge_scores = scorer.score(reference, [3])
print(f"BLEU: {bleu:.2f}")
print(f"ROUGE-1 F1: {rouge_scores['rouge1'].fmeasure:.2f}")

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

Acandidate

BRougeScorer

Cgenerated

Dreference

Attempts:

3 left

💡 Hint

Common Mistakes

Mixing up reference and generated variables.

Using lowercase 'rougescorer' instead of 'RougeScorer'.

The BLEU score uses the candidate tokens, the ROUGE scorer is created with RougeScorer, and ROUGE is scored comparing reference and generated text.