Machine Translation Mastery

Test

Parties jouées 21

Créer une copie

À propos de cette activité

MT basics and pitfalls

Créé par

ORALOVA

Uzbekistan

Téléchargez la version pour jouer sur papier

Top 10 résultats

#	Jeux	Précision	temps
1	Iroda Raxmonova 6 Mai 2026	% 70	00:45

Voulez-vous apparaître dans le Top 10 de ce jeu? Connectez-vous pour vous identifier.

Créez votre propre jeu gratuite à partir de notre créateur de jeu

Affrontez vos amis pour voir qui obtient le meilleur score dans ce jeu

Créer un défi

Practice Quiz - Punnett Square, Variation, and Mutations

Benjamin D Morgan

United States

(14)

Practice Quiz - Punnett Square, Variation, and Mutations
Tissues - Practice Quiz

Benjamin D Morgan

United States

(16)

Tissues - Practice Quiz
Photosynthesis vs Cellular Respiration - Practice Quiz

Benjamin D Morgan

United States

(39)

Photosynthesis vs Cellular Respiration - Practice Quiz
UPS 5 Seeing Habits Quiz

Heather Bell

United States

(12)

Test your knowledge about the 5 seeing habits discussed in the text 'UPS 5 seeing habits'.
Spanish Alphabet Listening

Falkenberg Falkenberg

United States

(88)

Spanish alphabet, listening-based assessment

		%
	Anonyme	%
		%
		%

Imprimez votre jeu

Machine Translation Mastery


Machine Translation MasteryVersion en ligne MT basics and pitfalls par ORALOVA 1 What is the primary difference between rule-based and neural machine translation (NMT)? a NMT ignores context and produces random text b Rule-based relies on bilingual dictionaries only c Rule-based uses word-by-word mapping only d NMT uses neural networks to model translations end-to-end 2 Which metric is commonly used to evaluate MT quality by comparing to reference translations? a BLEU score b METEOR score c ROUGE score d F1 score 3 Which challenge is MT most prone to due to polysemy and context? a Translating numerals b Detecting source language only c Choosing a meaning that fits the sentence d Identifying proper nouns universally e Correct spelling errors 4 What is a common risk when deploying MT in critical domains like healthcare or law? a Potential for inaccurate or harmful translations b Translations are always perfect c MT can only translate short phrases d MT provides legal/medical advice directly e MT eliminates need for human review 5 Which technique helps MT systems handle rare words or new terminology? a Subword tokenization (e.g., BPE) b Character-by-character translation only c Full word memorization d Ignoring unknown words and skipping them 6 What is beam search primarily used for in MT decoding? a Converting translations into numeric features b Exploring multiple translation hypotheses to find a better one c Reducing model size during inference d Training the model from scratch 7 Which issue can cause MT to produce fluent but semantically wrong translations? a Overreliance on surface form and fluency b Too slow processing time c Excessive use of proper nouns d Using too many synonyms e Lack of punctuation in source text 8 What is the role of a post-edit in MT workflows? a Generating synthetic data for training b Human corrections to improve translation quality c Replacing the MT model entirely d Translating text manually without MT 9 Which language feature can challenge MT due to syntactic differences between languages? a Word order and agreement rules b Color terminology c Monetary policy terms d Email address formats e Calendar time zones 10 Which scenario illustrates a bias or ethical concern in MT systems? a Unintentionally propagating stereotypes through translations b Skipping content due to censorship c Always translating literally with perfect accuracy d Always requiring user authentication before translation e Translating only neutral terms Feedback 1 Rule-based systems rely on linguistic rules and dictionaries; NMT learns from data end-to-end to capture context. 2 BLEU compares n-gram overlap with reference translations to estimate quality. 3 Context is essential; polysemous words require semantic fit within the sentence. 4 MT can misinterpret nuances; human post-editing is often necessary in critical domains. 5 Subword methods break rare terms into smaller units to improve coverage. 6 Beam search keeps several candidate translations to improve output quality. 7 Fluency can masquerade as accuracy if semantic alignment is missed. 8 Post-editing adjusts machine outputs to meet quality and domain requirements. 9 Different languages have different syntax; MT must align word order and agreement. 10 MT can reflect training data biases; careful evaluation is needed.

Vous pouvez intégrer le jeu dans un LMS compatible avec LTI 1.1 ou LTI 1.3 comme Canvas, Moodle ou Blackboard. Les scores seront ainsi automatiquement enregistrés dans le carnet de notes de la plateforme.

Machine Translation Mastery

Test

À propos de cette activité

Créé par

Téléchargez la version pour jouer sur papier

Top Jeux

Test

Practice Quiz - Punnett Square, Variation, and Mutations

Test

Tissues - Practice Quiz

Test

Photosynthesis vs Cellular Respiration - Practice Quiz

Test

UPS 5 Seeing Habits Quiz

Test

Spanish Alphabet Listening

Machine Translation MasteryVersion en ligne

par ORALOVA

What is the primary difference between rule-based and neural machine translation (NMT)?

Which metric is commonly used to evaluate MT quality by comparing to reference translations?

Which challenge is MT most prone to due to polysemy and context?

What is a common risk when deploying MT in critical domains like healthcare or law?

Which technique helps MT systems handle rare words or new terminology?

What is beam search primarily used for in MT decoding?

Which issue can cause MT to produce fluent but semantically wrong translations?

What is the role of a post-edit in MT workflows?

Which language feature can challenge MT due to syntactic differences between languages?

Which scenario illustrates a bias or ethical concern in MT systems?

Feedback