TMG301_SU26_FE_761954

1. (Choose 3 answers)
Which evaluation metrics are commonly used for machine translation?
A. BLEU score
B. METEOR score
C. TER metric
D. Monitor resolution
Đáp án:

2. (Choose 3 answers)
Which characteristics define modern neural dialogue systems?
A. Context-aware responses
B. End-to-end neural architectures
C. Large pretrained language models
D. Normalization data
Đáp án:

3. (Choose 3 answers)
Which algorithms are used for constituency parsing?
A. CYK algorithm
B. Earley parser
C. Chart parsing
D. Bubble parsing
Đáp án:

4. (Choose 3 answers)
Which characteristics define Transformer-based language models?
A. Self-attention mechanism
B. Contextual embeddings
C. Parallel sequence processing
D. Semantic vectors
Đáp án:

5. (Choose 1 answer)
What is the primary goal of relation extraction in text mining?
A. To extract entities and their attributes from text
B. To identify and categorize relationships between entities mentioned in text
C. To generate summaries of long documents
D. To translate text from one language to another
Đáp án:

6. (Choose 1 answer)
Which platform commonly hosts chatbot services?
A. Messaging applications
B. Disk utilities
C. BIOS interface
D. Compiler tools
Đáp án:

7. (Choose 1 answer)
What does a production rule in CFG define?
A. How symbols can be replaced to form valid sentences
B. How files are stored in memory
C. How GPUs process tensors
D. How networks route packets
Đáp án:

8. (Choose 1 answer)
What is the main goal of relation extraction in text mining?
A. Identify relationships between entities in text
B. Classify documents into predefined topics
C. Reduce vocabulary size in text preprocessing
D. Generate word embeddings for tokens
Đáp án:

9. (Choose 3 answers)
Which representations are often used as inputs to RNN models?
A. Word embeddings
B. Sentence embeddings
C. TF-IDF vectors
D. Normalization
Đáp án:

10. (Choose 1 answer)
Which type of temporal relation indicates that one event occurs immediately after another in a document?
A. Temporal precedence
B. Temporal overlap
C. Temporal succession
D. Temporal simultaneity
Đáp án:

11. (Choose 1 answer)
Which task allows chatbots to respond appropriately to user queries?
A. Intent detection
B. File compression
C. Hardware acceleration
D. Topic modeling
Đáp án:

12. (Choose 3 answers)
Which datasets support training event extraction systems?
A. ACE dataset
B. TAC KBP dataset
C. Annotated news corpora
D. VieM datasets
Đáp án:

13. (Choose 1 answer)
What is lemmatization in text normalization?
A. Simplifying words to their base or root forms
B. Removing special characters from text
C. Breaking down text into smaller units
D. Replacing numbers with words
Đáp án:

14. (Choose 1 answer)
What is the main idea of vector semantics in text mining?
A. Represent words as numerical vectors
B. Store text as images
C. Encrypt sentences
D. Compress documents
Đáp án:

15. (Choose 3 answers)
Which challenges occur when training RNN models?
A. Vanishing gradients
B. Exploding gradients
C. Long training time
D. Language modeling
Đáp án:

16. (Choose 3 answers)
Which features help detect temporal expressions?
A. Context words
B. Part-of-speech tags
C. Dependency relations
D. Words
Đáp án:

17. (Choose 1 answer)
What is fine-tuning in the context of pretrained language models?
A. Adapting a pretrained model to a specific task
B. Deleting training data
C. Compressing model weights
D. Increasing data
Đáp án:

18. (Choose 3 answers)
Which representations capture semantic meaning of words for modern text analysis tasks?
A. Word embeddings
B. Sentence embeddings
C. Contextual embeddings (e.g., BERT)
D. Text extraction
Đáp án:

19. (Choose 1 answer)
In first-order logic, what is the main distinction between a "function" and a "predicate"?
A. Functions are used to represent relations between entities, while predicates denote specific actions
B. Functions return entities, whereas predicates return truth values
C. Functions operate on quantifiers, while predicates operate on propositional logic
D. Functions are always true, while predicates can be false
Đáp án:

20. (Choose 1 answer)
What is a context-free grammar (CFG)?
A. A set of production rules used to describe the syntax of a language
B. A method for compressing text files
C. A database indexing technique
D. A neural network architecture
Đáp án:

21. (Choose 3 answers)
Which applications commonly rely on fine-tuned pretrained models?
A. Chatbots
B. Text summarization
C. Information extraction
D. Word embeddings
Đáp án:

22. (Choose 3 answers)
Which methods can improve robustness of text matching systems?
A. Edit distance
B. Regular expressions
C. Normalization
D. Word embeddings
Đáp án:

23. (Choose 1 answer)
What is an N-gram in text mining?
A. A sequence of N consecutive words or tokens
B. A type of database
C. A visualization technique
D. N-gram language modeling
Đáp án:

24. (Choose 1 answer)
Which N-gram consists of three consecutive tokens?
A. Trigram
B. Unigram
C. Bigram
D. Pentagram
Đáp án:

25. (Choose 3 answers)
Which inputs are required when training masked language models?
A. Tokenized text
B. Masked tokens
C. Large text corpus
D. Normalization
Đáp án:

26. (Choose 3 answers)
Which modeling assumptions apply to logistic regression?
A. Log-odds linearity
B. Independent observations
C. Probabilistic interpretation
D. Hard drive rotation
Đáp án:

27. (Choose 1 answer)
What does a pretrained language model learn from large corpora?
A. General linguistic patterns
B. Disk storage structure
C. Word embeddings tables
D. Normalization
Đáp án:

28. (Choose 1 answer)
What is the primary objective of the "masked language modeling" pre-training task?
A. To predict the masked words in a sentence
B. To translate text from one language to another
C. To generate new text based on a given context
Đáp án:

29. (Choose 3 answers)
Which datasets are commonly used to train dependency parsers?
A. Treebanks
B. Annotated corpora
C. Universal Dependencies datasets
D. KAM data
Đáp án:

30. (Choose 3 answers)
Which challenges occur in word sense disambiguation?
A. Context ambiguity
B. Sparse labeled data
C. Multiple valid interpretations
D. Text extraction
Đáp án:

31. (Choose 3 answers)
Which tasks are related to machine translation research?
A. Cross-lingual representation learning
B. Multilingual language modeling
C. Domain adaptation
D. N-gram language modeling updates
Đáp án:

32. (Choose 3 answers)
Which characteristics describe lexical semantic resources like WordNet?
A. Structured semantic relations
B. Synset-based representation
C. Concept hierarchy
D. GPU cooling
Đáp án:

33. (Choose 1 answer)
Which language model predicts the next word in a sequence?
A. N-gram model
B. Decision Tree
C. Naive Bayes
D. K-means
Đáp án:

34. (Choose 1 answer)
Which example is a temporal expression?
A. January 1, 2025
B. Artificial Intelligence
C. Hard disk
D. GPU
Đáp án:

35. (Choose 1 answer)
Which visualization shows the relationship between topics and documents?
A. Confusion matrix
B. t-SNE plot
C. Bar chart
D. Word cloud
Đáp án:

36. (Choose 1 answer)
In regular expressions, what does the "+" quantifier mean?
A. Matches zero or one occurrence
B. Matches zero or more occurrences
C. Matches one or more occurrences
D. Matches exactly three occurrences
Đáp án:

37. (Choose 1 answer)
What does "TF-IDF" stand for in the context of text clustering?
A. Term Frequency-Inverse Document Frequency
B. Text Feature Identification
C. Text File Description
Đáp án:

38. (Choose 3 answers)
Which algorithms are used for dependency parsing?
A. Transition-based parsing
B. Graph-based parsing
C. Neural dependency parsing
D. Bubble parsing
Đáp án:

39. (Choose 3 answers)
Which factors affect logistic regression generalization ability?
A. Regularization strength
B. Training data size
C. Feature engineering
D. Keyboard layout
Đáp án:

40. (Choose 3 answers)
Which NLP tasks use temporal reasoning?
A. Question answering
B. Information extraction
C. Event timeline generation
D. Topic modeling
Đáp án:

41. (Choose 3 answers)
Which preprocessing steps are used before training RNN models for text?
A. Tokenization
B. Vocabulary building
C. Embedding representation
D. Semantic embeddings
Đáp án:

42. (Choose 3 answers)
Which techniques are used to address data sparsity in N-gram models?
A. Smoothing
B. Backoff
C. Interpolation
D. Unicode normalization
Đáp án:

43. (Choose 1 answer)
In a transformer-based model, what is the role of the "multi-head attention mechanism"?
A. To attend to different heads of an organization
B. To enable the model to focus on multiple parts of the input sequence simultaneously, capturing different types of information
C. To reduce the number of parameters in the model
Đáp án:

44. (Choose 1 answer)
In text mining, logistic regression is commonly used for?
A. Text classification
B. File compression
C. Video processing
D. Hardware design
Đáp án:

45. (Choose 3 answers)
Which improvements can enhance large-scale document clustering?
A. Distributed computing
B. Efficient indexing
C. Dimensionality reduction
D. Semantic embeddings
Đáp án:

46. (Choose 3 answers)
Which components are part of logical sentence representations?
A. Predicates
B. Variables
C. Logical operators
D. Part-of-speech tags
Đáp án:

47. (Choose 1 answer)
In sentiment analysis, what does a lexicon entry typically consist of?
A. A list of related words with their part-of-speech tags
B. A numerical score representing the polarity or sentiment strength of a word
C. A definition of the word's meaning
D. The frequency of the word's occurrence in a given text
Đáp án:

48. (Choose 1 answer)
Which model type is commonly used in modern machine translation systems?
A. Transformer models
B. Decision Trees
C. K-Means clustering
D. Apriori algorithm
Đáp án:

49. (Choose 1 answer)
In event extraction, what does an "event trigger" refer to?
A. The time at which an event occurred
B. A word or phrase that denotes the occurrence of an event or action
C. A type of passive voice construction
D. A type of prepositional phrase
Đáp án:

50. (Choose 1 answer)
Which of the following word embedding models is capable of handling subword information?
A. Word2Vec
B. FastText
C. GloVe
D. LDA
Đáp án:

Đang tải dữ liệu

Đang tải dữ liệu

Đang tải dữ liệu