CAT 2020 Slot 3 VARC Q19
- Verbal Ability and Reading Comprehension
- Verbal Ability › Odd One Out
- Hard
- Type in the answer (TITA)
- CAT_2020_SLOT3_VARC_Q19
Question 19
Five jumbled up sentences (labelled 1, 2, 3, 4 and 5), related to a topic, are given below. Four of them can be put together to form a coherent paragraph. Identify the odd sentence and key in the number of that sentence as your answer.
-
Machine learning models are prone to learning human-like biases from the training data that feeds these algorithms.
-
Hate speech detection is part of the on-going effort against oppressive and abusive language on social media.
-
The current automatic detection models miss out on something vital: context.
-
It uses complex algorithms to flag racist or violent speech faster and better than human beings alone.
-
For instance, algorithms struggle to determine if group identifiers like "gay" or "black" are used in offensive or prejudiced ways because they're trained on imbalanced datasets with unusually high rates of hate speech.
Type-in-the-answer (TITA) question: no options are given; enter the value.
Show answer and solution
Answer:
3
Solution
🎯 Question Analysis & Solution
-
Question Type: Odd Sentence Out (TITA)
-
Passage Reference:
Sentence 2: "Hate speech detection is part of the on-going effort against oppressive and abusive language on social media."
Sentence 4: "It uses complex algorithms to flag racist or violent speech faster and better than human beings alone."
Sentence 1: "Machine learning models are prone to learning human-like biases from the training data that feeds these algorithms."
Sentence 5: "For instance, algorithms struggle to determine if group identifiers like 'gay' or 'black' are used in offensive or prejudiced ways because they're trained on imbalanced datasets with unusually high rates of hate speech."
Sentence 3: "The current automatic detection models miss out on something vital: context."
Logical Deduction
-
Sentence 2 establishes the macro-subject: automated hate speech detection on social platforms.
-
Sentence 4 directly follows Sentence 2 through pronominal reference ("It"), explaining the operational speed and complexity of the algorithms used.
-
Sentence 1 introduces the specific technical vulnerability: algorithmic bias resulting from biased training data.
-
Sentence 5 provides an explicit case in point ("For instance"), detailing how imbalanced datasets cause algorithms to misclassify group identifiers like "gay" or "black."
-
Together, Sentences 2-4-1-5 form an integrated thematic paragraph tracing hate speech detection to dataset-induced algorithmic bias.
-
In contrast, Sentence 3 introduces a distinct linguistic and computational issue—the absence of "context"—diverging from the dataset bias narrative developed across the other four sentences.
Correct Answer: 3
- Sentence 3 is the odd sentence out because it introduces the separate linguistic issue of semantic context, whereas Sentences 2, 4, 1, and 5 form a unified structural argument centered exclusively on dataset-induced algorithmic bias.
Option Elimination / Coherent Group Analysis
-
Coherent Group (2-4-1-5): Sentence 2 introduces hate speech detection; Sentence 4 describes its algorithmic execution; Sentence 1 pinpoints its vulnerability to training dataset bias; and Sentence 5 provides a concrete example of this training data imbalance.
-
Sentence 3 (Out of Scope / Topical Divergence): Diverges from the coherent theme by shifting the critique to the lack of "context," an independent natural language processing defect that breaks the linear focus on training dataset bias.
Detailed explanation
📖 Structural Flow & Logical Progression
Central Theme
The structural mechanisms and inherent algorithmic biases in automated hate speech detection models, specifically examining how imbalanced training datasets cause algorithms to adopt human-like prejudices regarding demographic identifiers.
Key Logical Connectors & Structural Flow
-
Introduction of Subject (Sentence 2): Introduces automated hate speech detection as an instrument targeting abusive online discourse.
-
Mechanism of Action (Sentence 4): Connects via the pronoun "It" to hate speech detection, explaining its algorithmic operation in flagging hostile content faster than humans.
-
Systemic Problem / Complication (Sentence 1): Introduces the core algorithmic defect: machine learning models absorb human biases from their underlying training data.
-
Empirical Exemplification (Sentence 5): Direct illustration ("For instance") showing how models trained on skewed datasets struggle to parse identifiers like "gay" or "black."
Tone and Perspective
- Tone: Objective, technical, and analytical.
Odd One Out questions from other CAT papers
- CAT 2025 Slot 1 VARC Q1
- CAT 2025 Slot 1 VARC Q23
- CAT 2025 Slot 2 VARC Q3
- CAT 2025 Slot 2 VARC Q4
- CAT 2025 Slot 3 VARC Q4
Practise this question with timing, notes and your mistake notebook.
Practise in PrepSoc