Text Analytics DATA-SCIENCE-FOUNDATIONS Practice Questions (Page 3)

Questions 11–15

1. 11foundation · easy

   In a bag-of-words model, how is a document represented?

   • AAs a list of unique words in the document with their part-of-speech tags
   • BAs a vector of word counts, ignoring word order and grammar
   • CAs a sequence of word embeddings that preserve semantic relationships
   • DAs a tree structure that represents the grammatical relationships between words
   Reveal Solution Discussion Select an answer first
2. 12foundation · easy

   When interpreting the results of a topic modeling analysis, what does a high probability of a topic in a document indicate?

   • AThe document contains only words from that topic and no others
   • BThe document is definitely misclassified and should be removed
   • CThe document is the most important document in the corpus
   • DThe document is strongly associated with that topic
   Reveal Solution Discussion Select an answer first
4. 13application · medium

   A bank wants to automatically classify loan applications as 'approved' or 'denied' based on the applicant's written statements. They have 10,000 labeled applications. Which steps should they take to build and evaluate this text classification model?

   • AUse sentiment analysis to determine if the application is positive or negative
   • BPreprocess text, convert to TF-IDF features, train a classifier, and evaluate on a held-out test set
   • CApply LDA to discover topics and use them as the classification labels
   • DExtract named entities and use them as the only features without preprocessing
   Reveal Solution Discussion Select an answer first
5. 14expert · hard

   A marketing team ran LDA topic modeling on customer survey responses and found a topic with high-weight words: 'price', 'expensive', 'cost', and 'affordable'. They also ran sentiment analysis and found that responses containing these words have an average sentiment score of -0.7 (on a scale from -1 to 1). Which business insight is most strongly supported by these results?

   • ACustomers are concerned about pricing, and the negative sentiment suggests dissatisfaction with the cost
   • BCustomers are neutral about the product
   • CCustomers are satisfied with the product quality
   • DCustomers want more product features
   Reveal Solution Discussion Select an answer first
6. 15application · medium

   A data analyst is preparing a corpus of product reviews for topic modeling. The reviews contain many misspellings and informal abbreviations like 'gr8' and 'luv'. Which preprocessing step is most important to improve the quality of the topic model?

   • ANormalize the text by correcting misspellings and expanding abbreviations
   • BRemove stop words only
   • CSkip preprocessing and use raw text directly
   • DApply stemming only
   Reveal Solution Discussion Select an answer first