Machine Translation

With what metric are translation systems often evaluated?

• GRIS

• VERT

• BLEU

• ROSE

How can the BLEU score be calculated?

• sentence length

• n-gram counts (/comparisons)

• PROPN counts (/comparisons)

• word frequency

Translation errors are more easily traced back to their root in ...

• Neural MT systems

• Statistical MT systems

• Flawless systems

• Rule-based MT systems

On which level do we usually not align texts?

• Word level

• Character level
• Sentence level

• Document level

A commonly used approach nowadays for NMT models is the ... architecture.

• decoder-encoder

• ReLU

• encoder-decoder

• Latent Dirichlet allocation

Which of the following aspects do NMT systems sometimes struggle with?

• repetitions

• word choice

• simple spelling mistakes

• some word order errors

What is the fundamental task of many NLP systems?

• reading

• inferring

• sorting

• predicting

How many Bigrams are in the sentence (ignore punctuation and there are no start of sentence or end of sentence tags to be regarded): "I am sorry, I wrote this presentation and Quizizz at 3am"

• 10

• 5

• 12

• 9

What is the TF of the word "crying" in the 2 documents:
Doc1: "Today I was crying crying in the bathroom"
Doc2: "Today you were crying"

• Doc1: 1/8
  Doc2: 1/4

• Doc1: 2/3
  Doc2: 1/4

• Doc1: 2/8
  Doc2: 1/4

• Doc1: 2/8
  Doc2: 2/4