Task 1.3: Implement Data Validation and Processing Pipelines for FM Consumption. AIP-C01 Practice Questions (Page 4)

Questions 16–20

1. 16expert · hard

   A company is building a real-time customer support chatbot using Amazon Bedrock. The chatbot uses a RAG architecture with a vector database. The pipeline ingests support documents from various sources (PDFs, Word docs, HTML pages) into the vector database. The team has noticed that the chatbot's responses are sometimes inaccurate because the ingested documents contain outdated information and duplicate content. They need to implement a data validation and processing pipeline to ensure the vector database only contains accurate, unique, and current information. The pipeline must also be able to handle updates to the source documents. What is the MOST comprehensive approach?

   • AUse an event-driven pipeline with Lambda functions triggered by S3 events. Use a Lambda function to check for duplicates in the vector database, and use another to validate the document's last-modified date against a metadata store.
   • BUse Amazon Comprehend to analyze the sentiment of the documents and filter out any documents with negative sentiment.
   • CUse a SageMaker Processing job to re-embed all documents and use a custom script to compare the new embeddings with the old ones to find duplicates.
   • DUse a scheduled AWS Glue ETL job to process all documents, use Glue Data Quality to check for duplicates and outdated dates, and then overwrite the entire vector database index.
   Reveal Solution Discussion Select an answer first
2. 17foundation · easy

   A company is building a multimodal application that accepts both images and text as input and generates a response that combines both modalities. Which AWS service provides foundation models that can natively process both image and text inputs in a single API call?

   • AAmazon Bedrock
   • BAmazon SageMaker Processing
   • CAWS Lambda
   • DAmazon Rekognition
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   A legal firm is using Amazon Bedrock to analyze contracts. They have a large number of PDF contracts that contain a mix of typed text and handwritten notes. The handwritten notes are often critical for understanding the contract's intent. They want to ensure that the text extracted from these PDFs is complete and accurate before sending it to the FM for analysis. Which service should they use for the initial text extraction?

   • AAmazon Transcribe, because it can convert the audio from a dictation of the contract into text.
   • BAmazon Comprehend, because it can extract entities and key phrases from the text.
   • CAmazon Rekognition, because it can detect objects and scenes in the document images.
   • DAmazon Textract, because it can detect and extract text from both printed and handwritten documents.
   Reveal Solution Discussion Select an answer first
5. 19foundation · easy

   A data engineer is designing a validation workflow for a dataset that will be used to fine-tune a foundation model. The workflow must automatically check for common data quality issues such as missing values, duplicate rows, and schema mismatches, and then publish metrics to Amazon CloudWatch. Which AWS service is purpose-built to define and run these data quality checks on data stored in AWS Glue Data Catalog tables?

   • AAWS Glue Data Quality
   • BAmazon SageMaker Data Wrangler
   • CAmazon CloudWatch Evidently
   • DAWS Lambda custom function
   Reveal Solution Discussion Select an answer first
6. 20application · medium

   A company is building a system to automatically generate meeting summaries. The system records audio from meetings, transcribes it, and then uses Amazon Bedrock to generate a summary. The team wants to improve the quality of the summaries by providing the FM with additional context, such as the names of the participants and the main topics discussed. What is the most effective way to generate this additional context?

   • AUse Amazon Transcribe to transcribe the audio, and then use Amazon Comprehend to extract the key entities (participant names) and key phrases (topics) from the transcript.
   • BUse Amazon Rekognition to analyze the video of the meeting to identify the participants and their names.
   • CUse Amazon Bedrock to directly analyze the raw audio file and generate a summary with participant names and topics.
   • DUse Amazon Polly to convert the audio to text and then use a Lambda function to search for common names and topics.
   Reveal Solution Discussion Select an answer first