More on MT

What is the bigram precision of the following reference-hypothesis pair?

What is the advantage of the modified n-gram precision?

• it is easier to calculate

• very short hypotheses are penalized

• repetitions of correct n-grams aren't counted too often

• it can modify other evaluation measurements

What are the unigram and the modified unigram precisions for this sentence?

On which level do we usually not align texts?

• words

• document

• genre

• sentence

What does this diagram show?

• Convolutional Neural Network

• Recurrent Neural Network

• TnT Tokenizer

• Feed-Forward Neural Network

What advantage does SMT have over NMT?

• there are no advantages

• shorter training time

• more fluent output

• less computational power required

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

• decoder-encoder

• ReLU

• encoder-decoder

• FFNN

Which of these do NMT systems sometimes struggle with?

• some word order errors

• word choice

• simple spelling mistakes

• consistency across sentences

When did the era of NMT start?

• 2019/2020

• 2014/2015

• 1990s

• 2005/2006

What is the fundamental task of many NLP systems?

• sorting

• inferring

• reading

• predicting