Unit 1: Foundations of NLP and Text Processing - Practice Quiz

1 What does NLP primarily combine to help computers process human language?

2 What is grammar in a language?

3 Which linguistic level focuses on the smallest units of sound in spoken language?

4 What does morphology study in linguistics?

5 What does syntax describe?

6 What is the main focus of semantics?

7 Why can ambiguity make NLP difficult?

8 Which is a common application of NLP?

9 What is tokenization?

10 What is the goal of stemming?

11 How does lemmatization differ from stemming?

12 What are stop words?

13 Why might punctuation be removed during text preprocessing?

14 What is an out-of-vocabulary word?

15 Which is an example of text normalization?

16 What information does a Bag-of-Words representation mainly record?

17 What is a bigram?

18 What does TF-IDF help measure in a document?

19 In TF-IDF, what does the term IDF represent?

20 In the word "unhappiness," which part is a suffix?

21 An early machine translation system replaces each source-language word with a dictionary equivalent but produces sentences with incorrect word order. Which historical limitation of early NLP does this best illustrate?

22 A grammar accepts "The students solve the problem" but rejects "The students solves the problem." Which grammatical constraint is being enforced?

23 In the sentence "They can book the tickets," a part-of-speech tagger labels "book" as a verb. Which contextual clue most directly supports this label?

24 Consider the word "unhappiness." Which segmentation best represents its morphemes?

25 A parser analyzes "The scientist saw the assistant with the telescope." What syntactic issue must it resolve?

26 A system must determine whether "bank" refers to a financial institution or the side of a river. Which NLP task is most directly required?

27 A sentiment classifier labels "Great, another three-hour delay" as positive because of the word "Great." Which NLP challenge caused the error?

28 A hospital wants to identify medication names, dosages, and frequencies in clinical notes. Which NLP application best fits this requirement?

29 A tokenizer splits "don't" into "do" and "n't" instead of keeping it as one token. What is the main benefit of this choice for many English NLP models?

30 A social-media tokenizer processes "Email me at ana@example.com!" Which output is most useful when email addresses should remain meaningful units?

31 A stemmer maps both "university" and "universe" to the same shortened form, reducing search precision. What property of stemming explains this behavior?

32 A preprocessing system should map "better" to "good" and "cars" to "car." Why is lemmatization more suitable than basic stemming?

33 A sentiment model removes common stop words and then confuses "I recommend this" with "I do not recommend this." What is the best preprocessing adjustment?

34 A sentiment system processes "The service was good!!!" What is the most appropriate treatment of the exclamation marks?

35 A model encounters the unseen word "microblogging" but its vocabulary contains subwords "micro," "blog," and "ing." Which strategy best avoids representing the word as a single unknown token?

36 A search engine should match documents containing "N.L.P.," "nlp," and "NLP" when a user searches for "nlp." Which normalization pipeline is most appropriate?

37 Using the vocabulary [deep, learning, models], what is the Bag-of-Words vector for "learning models learning"?

38 Which representation can distinguish "not useful" from "useful" while still using a simple count-based model?

39 After tokenizing "natural language processing is useful" into five words, how many contiguous word trigrams can be extracted without padding?

40 Two terms occur five times each in a document. The first appears in nearly every document in the corpus, while the second appears in only two documents. Under TF-IDF, which term usually receives the larger weight?

41 Why did early rule-based NLP systems often perform well in narrowly defined domains but fail when transferred to general language understanding tasks?

42 A grammar accepts the sentence "The scientist observed the student with the telescope" with two different parse structures. Which property of natural language does this example primarily demonstrate?

43 A sentiment classifier labels "The movie is hardly good" as positive because it detects the word "good." Which linguistic capability is most directly missing?

44 Which analysis best explains why a morphologically rich language can create severe data sparsity for a word-based NLP model?

45 Why can a bag-of-words representation fail to distinguish "The dog chased the cat" from "The cat chased the dog"?

46 A system translates "bank" incorrectly because it always selects the financial meaning, even when the surrounding text discusses rivers. What semantic problem is illustrated?

47 A model trained on news articles performs poorly on medical forum posts despite similar vocabulary. Which factor is the most likely primary cause?

48 Which NLP application requires modeling both the factual content of a document and the user's information need to rank results effectively?

49 Why is whitespace tokenization inadequate for processing the text "Email me at data.science@example.com"?

50 For a sentiment task, why might splitting "can't" into "can" and "n't" be preferable to treating it as one token?

51 A stemmer maps "university" and "universe" to the same stem. What is the principal risk of this behavior?

52 Why does lemmatization generally require more linguistic information than stemming?

53 Why can removing stop words harm a question-answering system that processes the query "Who did not attend?"?

54 A tokenizer removes all punctuation before sentiment classification. Which input is most likely to lose important information as a result?

55 A word-level model maps every unseen word to one <UNK> token. Why can this be inadequate for recognizing a new medical term such as "cardioneurography"?

56 Which normalization strategy is most appropriate when matching user-entered product names while preserving distinctions that may affect product identity?

57 Two documents have identical word counts but opposite meanings because one uses "not effective" and the other uses "effective, not" in a task where word order matters. What limitation is being exposed?

58 What is the main trade-off when increasing an NLP model's representation from unigrams to trigrams?

59 Suppose a term occurs in every document of a corpus. Under standard inverse document frequency, how should its IDF contribution compare with that of a term appearing in only one document?

60 Using , what is the TF-IDF weight of a term appearing 3 times in one document when and ?