Natural Language Processing (NLP) AIE Practice Questions (Page 4)

Questions 16–20

1. 16foundation · easy

   Which of the following is a common challenge in NLP related to the same word having multiple meanings depending on context?

   • ALexical ambiguity
   • BStop-word removal
   • CPart-of-speech tagging
   • DTokenization
   Reveal Solution Discussion Select an answer first
2. 17expert · hard

   A team is building a language model for a specialized domain (medical records). They have a large corpus of de-identified medical text but limited computational resources. They are considering: (1) training a transformer model from scratch, (2) fine-tuning a pre-trained general language model, (3) using a statistical n-gram model, (4) using a simple recurrent neural network. Which approach is most likely to achieve good performance with limited computational resources?

   • ATrain a transformer model from scratch on the medical corpus.
   • BUse a simple recurrent neural network.
   • CUse a statistical n-gram model.
   • DFine-tune a pre-trained general language model on the medical corpus.
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   A legal firm wants to use NLP to extract clauses from contracts. The same word can have different meanings depending on the context (e.g., 'party' as a social gathering vs. a legal entity). Which approach would best help the model disambiguate such terms?

   • AUse a language model that incorporates surrounding context to infer word meaning.
   • BApply stemming to reduce all words to their root forms.
   • CRemove all stop words from the contract text.
   • DUse a simple keyword-matching algorithm to find the word 'party'.
   Reveal Solution Discussion Select an answer first
5. 19application · medium

   A global news agency wants to automatically classify articles as positive, negative, or neutral. They notice that the phrase 'this is sick' is labeled negative in US English but positive in UK English slang. Which NLP challenge is the system facing?

   • ACultural and contextual nuance
   • BNamed entity recognition failure
   • CTokenization ambiguity
   • DPart-of-speech tagging errors
   Reveal Solution Discussion Select an answer first
6. 20application · medium

   A team is developing a next-word prediction model for a search engine. They want the model to understand that 'bank' can mean a financial institution or a river bank. Which type of language model is best suited for this?

   • AA contextual neural language model like BERT.
   • BA simple n-gram model with a fixed window.
   • CA bag-of-words model.
   • DA rule-based system with keyword matching.
   Reveal Solution Discussion Select an answer first