SBL1400 Lecture 2 Quiz – Text & Sentiment Analysis

What makes unstructured data such as text, audio, and video harder to analyze than structured data?

• It lacks visual components

• It changes in real-time

• It cannot be directly stored in tabular rows and columns

• It is typically free from noise

Which elements are essential when preprocessing textual data before analysis? Select all that apply.

• Tokenisation

• Lowercasing text

• Removing punctuation

• Converting to audio

• Stemming

The process of reducing words to their root forms in order to treat similar words as one is known as __________.

Tokenisation is the process of mapping text documents into rows and columns for numerical analysis.

• True
• False

Answer explanation

Explanation: That describes the term-document matrix; tokenisation splits text into individual terms.

Match each concept to its most appropriate definition:

• Numerical representation of emotional to

• Sentiment Score

• Contextual Interpretation

• Corpus

• Collection of text documents

• Term-Document Matrix

• Table showing presence/absence of terms

• Contextual Interpretation

A text analyzer identifies the sentence "My delayed flight was absolutely amazing!" as positive. What is the likely cause of this misinterpretation?

• Lack of punctuation

• Presence of ambiguous vocabulary

• Sarcasm misclassified as genuine sentiment

• Irrelevant tokens

Arrange the following preprocessing steps in logical order for text mining:​ (a)   ​ ​ (b)   ​ ​ (c)   ​ (d)  

• Convert to lowercase
• Remove punctuation
• Tokenize
• Apply stemming

In a term-document matrix, what does a column represent?

• A complete sentence

• A tokenised phrase

• A numerical variable

• A specific term or word across documents

High-frequency words always carry high semantic value and should be retained for analysis.

• True
• False

Answer explanation

Explanation: High-frequency words (e.g., “the”, “and”) may have low information value.

Why might synonym consolidation be necessary in text mining?

• To increase matrix dimensions

• To reduce redundancy and improve interpretability

• To inflate sentiment scores

• To align all documents to a single topic