Answer: Use Amazon Comprehend toxicity detection.

## Explanation **Correct Answer: B - Use Amazon Comprehend toxicity detection.** **Why this is correct:** 1. **Amazon Comprehend** is Amazon's natural language processing (NLP) service that includes a **Toxicity Detection** feature specifically designed to identify harmful language, hate speech, and toxic content. 2. **No labeled data required**: Amazon Comprehend's toxicity detection is a pre-trained model that works out-of-the-box without requiring any labeled training data from the user. 3. **Perfect fit for the use case**: The service is specifically designed for analyzing text content like social media comments to identify harmful language. **Why other options are incorrect:** **A. Use Amazon Rekognition moderation.** - Amazon Rekognition is primarily for **image and video analysis**, not text analysis. - While it has content moderation features, they are focused on visual content (inappropriate images/videos), not text comments. **C. Use Amazon SageMaker built-in algorithms to train the model.** - This would require **labeled data** to train the model, which contradicts the requirement that "the company will not use labeled data to train the model." - SageMaker is for building custom ML models, which typically require training data. **D. Use Amazon Polly to monitor comments.** - Amazon Polly is a **text-to-speech** service that converts text into lifelike speech. - It has no capability for content analysis, toxicity detection, or harmful language identification. **Key AWS Service Distinctions:** - **Amazon Comprehend**: NLP service for text analysis (sentiment, entities, key phrases, toxicity) - **Amazon Rekognition**: Computer vision service for image/video analysis - **Amazon SageMaker**: ML platform for building, training, and deploying custom models - **Amazon Polly**: Text-to-speech service The requirement for "no labeled data" makes Amazon Comprehend's pre-trained toxicity detection the ideal solution.

Author: Ritesh Yadav